저자: Alex · 외국인
요약 보기
엔비디아 실적 발표를 앞두고 AI 칩 밸류체인의 수혜 기업(ASML, 어플라이드머티리얼즈, TSMC, 마이크론, 슈퍼마이크로)을 소개하며 간접 투자 기회를 강조. 시장 점유율·마진·HBM 성장률 등 수치를 근거로 제시. 슈퍼마이크로는 10:1 액면분할 예정이라 긍정적으로 언급. 중간에 VPN 스폰서 광고 포함.
다듬은 글
Nvidia is the most powerful AI company on Earth because they make the most powerful AI chips on earth almost all AI workloads run on Nvidia gpus today but with their earnings right around the corner the real opportunity for investors is more than just Nvidia stock it's the other companies winning big by working with Nvidia but that almost nobody else is covering making them a great way are the companies I'm going to cover in this video asml which has the single biggest Monopoly in the entire AI Market Applied Materials and their Innovations in the semiconductor space Taiwan semiconductor which makes over 90% of the world's most advanced chips including nvidia's gpus Micron Technology which makes the high bandwidth memory modules for nvidia's black oil chips and super micro computer a data center infrastructure company that integrates those chips into massive data centers and has a 10 for one stock split on the way like I always say the best way to find great stocks before everyone else is to understand a company's products not just their profits huge growth happens when a company has the perfect product for a hot new market Nvidia is dominating data center gpus with a massive 92% market share and the market for data center accelerators in general is expected to more than 5x by 2032 which means demand for these chips is growing at a 24% compound annual growth rate or roughly twice as fast as the S&P 500 combine that with Nvidia having gross margins north of 78% which is as high as most software companies and it's easy to see why Nvidia stock is up by almost 200% over the last year and what all these companies in this video have in common is their involvement in making and deploying nvidia's chips let's start with asml asml is the only company in the world that makes extreme ultraviolet lithography machines these massive machines use ultraviolet light to etch the tiny complex chip designs on to Silicon Wafers each euv machine has more than 100,000 Parts but asml actually only makes around 10% of the parts in these machines the other 90% are made by hundreds of companies all around the world and asml acts as the systems integrator that way asml doesn't need to be the best in the world at making every single part many of which are highly specialized and require insane amounts of engineering as the name implies extreme ultraviolet light has an extremely short wavelength which is what lets asml machines print incredibly tiny details onto microchips the euv light is Created from Tiny explosions of molten tin that happen 50,000 times per second in each machine each pulse of light is then bounced off special mirrors made by Zeiss before reaching the Silicon wafer these are the flattest mirrors on Earth they're so flat that if Zeiss could make a mirror the size of the United States the biggest bump on the entire mirror would be the size of a single spec of dust and that's just one of over 100,000 Parts in these machines so it would be much riskier if asml did try to build every component themselves because if any one part has an issue the entire $200 million machine has an issue that's why they built a massive network of exclusive deals with hundreds of special suppliers experts say that it could take decades and billions of dollars for any other company to challenge asml moat around these machines which is why I said they have the biggest Monopoly in the entire AI market and their next Generation machine is called the high euv lithography machine which can etch lines up to 1.7 times thinner enabling up to 2.9 times more transistors per chip more transistors per chip means higher processing speeds more memory density and ultimately more compute per watt exactly what's in high demand thanks to the AI Revolution and the explosion of data that comes with it the amount of data on the internet has been growing at around 23% per year since 2021 and that doesn't account for text audio or video data created by generative Ai and speaking of data you probably know that vpns protect you and your data by encrypting it and making it look like you're somewhere else but did you know that VPN can also save you a a ton of money that's where surf shark comes in the sponsor of this video surf shark is a super fast and easy to use VPN that you can install on unlimited devices with just one account letting you protect yourself and your entire family and since some online stores raise their prices based on your location you can use surf shark to always get the best price on expensive purchases like plane tickets hotel rooms and rental cars not to mention that surf shark lets you access content websites and services that are blocked in your region altogether since vpns change your digital location so if you care about your family's privacy and you like saving money you can use my code ticker to get up to four extra months for free and since surf shark offers a 30-day money back garantee there's no risk in trying it with my link in the description below today and a big thank you to Surf shark and to you for supporting the Channel all right since asml has Monopoly on manufacturing these lithography machines they're going to benefit in a big way from the increase in demand for high performance and energy efficient AI chips even after the most recent downturn in tech stocks asml stock is still up by around 30% year-to date and I think they have plenty more upside over the long term next up is Applied Materials ticker symbol Amat Jensen Wong made it clear that nvidia's gpus and accelerated Computing in general would not exist without Applied Materials Applied Materials has revolutionized the semiconductor indust industry by inventing fundamental chip processing tools for example they make machines called epitaxial reactors which apply different layers of materials onto a semiconductor to change its electrical properties like making certain areas more or less conductive before it becomes a memory device or microprocessor Amat also makes chemical mechanical polishers or cmps polishing these layers helps them fit together without any gaps or electrical issues once all the layers have been added to the wafer by the reactor and Polished by the CMP it's considered a planarized wafer and it's ready for the next processing step which is etching and deposition at a high level etching is where the tiny channels are carved into each layer of the wafer the pattern these channels create is determined by the design of the circuit which comes from Chip designers like Nvidia after etching the deposition process is where a conductive metal is deposited into the channels turning the planarized wafer into an electronic circuit Applied Materials makes many more machines for a wide variety of Industries but I wanted to at least highlight a few of the big ones for chip design since this video is all about companies that support nvidia's gpus they've also talked about the huge demand for some of their other semiconductor Solutions on recent earnings calls for example emat expects their revenues from high bandwidth memory packaging to quadruple year over-year thanks to generative AI that means their revenues should keep expanding for years to come Amat stock is currently up by around 30% year-to date and just like asml stock I think Amat still has a lot of room to run over the long term all right so asml builds the world's most advanced lithography machines and Applied Materials makes more machines involved in chip fabrication the Taiwan semiconductor manufacturing company is where it all comes together tsmc is a Pure Play Chip Foundry that supports chip designers like Nvidia tsmc doesn't make any chips for themselves which means they never compete with their clients as a result they make around 60% of all chips on Earth and a whopping 90% of all advanced microchips like Apple's a series and M series chips for iPhones and MacBooks Tesla's chips for full self-driving and of course nvidia's gpus for data centers and high performance PCS in fact tsmc has a custom 4 nanometer process specifically for nvidia's chips called 4np which is currently set up to manufacture Blackwell gpus compared to their current generation of data center gpus called Hopper these new Blackwell chips are four times better at AI training and a whopping 30 times better at AI inference this massive leap in performance per watt is why nvidia's chips have so much demand compared to their competitors demand that flows down the supply chain to companies like tsmc since they're the only company on Earth with the ability to make these chips in high volumes and that's important because Blackwell actually combines two separate dieses onto a single chip and connects them with a 10 tbte pers second link so they act as one GPU two dies that each push the limits of asml machines plus an extra high-speed link plus a larger total area for advanced packaging means that Blackwell takes over twice the production capacity and Manufacturing resources as Hopper these kinds of Supply constraints are one of the reasons that nvidia's gpus cost so much in the first place tsmc's 3 nanometer process node N3 is currently in volume production and it's expected that nvidia's next generation of gpus after Blackwell called Ruben will be made on the N3 process node TSM stock is up by over 60% since the start of the year and over 80% since I started seriously covering it because of their close relationship with Nvidia and here's a fun fact tsmc uses nvidia's AI to make its chip manufacturing process more efficient Nvidia created a tool called autod DMP which tsmc uses to optimize the placement of Chip components up to 30 times faster than previous methods tsmc also uses nvidia's kitho library to accelerate computational lithography by 40 to 60x which is a huge breakthrough because that's a huge bottleneck in how fast modern chips can be designed that means tsmc is using nvidia's AI to build better chips those same chips make nvidia's AI more powerful which helps tsmc build even better chips so this is AI building Ai and we can't talk about AI without talking about memory Micron Technology ticker symbol mu is an American company that makes memory for the entire Computing industry not just Nvidia Micron makes solid state drives Ram modules and flash memory for data centers consumer PCS smartphones and even Automotive all of these markets are transitioning to more advanced memory because of AI applications from co-pilots on PCS and smartphones to AI on servers and in self-driving cars and because almost every kind of device will need more memory the market for high bandwidth memory is expected to almost 50x over the next 7 years which would be an insane 68% compound annual growth rate for the rest of the decade now to be fair these growth numbers are so big because High bandwith memory only makes up 5% of all memory deployed today while demand for AI Hardware is at an all-time high so the AI Hardware space is set for a ton of growth growth that's being dominated by Nvidia well Micron makes the high bandwidth memory chips for nvidia's Blackwell gpus high bandwidth memory packs more memory into a smaller space by stacking memory chips and connecting them all to a single Hub doing that lowers the physical distance that the data needs to travel which increases data transfer speeds while lowering overall power consumption Micron makes high band with memory cubes which use around 30% less power than their competition and just like Nvidia micron's most advanced hbm products have much more demand than Supply Micron has around a 25% share of the nextg memory Market putting them right behind Samsung and SK hyx both of which are headquartered in South Korea Micron stock is up by around 25% for the year even after going down by over 40% in the recent selloff in tech stocks as companies like Google meta platforms Microsoft and open AI keep building more advanced AI models they'll need tons more high bandwidth memory and better data storage solutions so no matter who wins Micron wins making it a great stock to get rich without getting lucky and that brings us right to Super Micro computer ticker symbol smci smci was founded in Silicon Valley in 1993 just like Nvidia super micro designs and integrates everything from fullscale server systems down to individual workstations for a wide variety of markets from cloud services and AI training to 5G and Edge Computing but instead of competing with companies like Nvidia and AMD they partner with them and customize their Computing Solutions adding the right high-speed networking Solutions Power Electronics and Cooling and even the right firmware and software as needed super micro has two huge competitive advantages first their server Solutions are completely modular super micro calls their modules server building blocks kind of like Lego pieces but for data centers some customers might need more memory While others might want more storage some data centers might want more CPUs While others want to deploy as many accelerators as they can super micro also has building blocks for Power Cooling and networking that can be optimized for specific data centers and specific workloads this building block approach is what lets smci quickly build solutions for so many different markets when a company like Nvidia launches a new chip super micro doesn't need to design a whole new server system for it they just design a new building block that connects to the rest of their ecosystem the other big Advantage is being headquartered in Silicon Valley not Taiwan or China that means when Nvidia designs a new GPU like Blackwell Nvidia can share the chip designs with super micro to start laying out and optimizing the boards the connectors and Cooling and everything else that keeps the B100 chips running at Peak Performance once they're installed in actual data centers and because they're right down the street updates and changes to the Chip's design are easier to share and coordinate this is a serious Advantage Nvidia recently delayed their Blackwell gpus by 3 months due to design flaws so as those designs evolve Nvidia will want super micro to minimize Blackwell's time to Market so that their customers can start earning Returns on their billion dooll infrastructure Investments faster especially if that means Nvidia can get their chips up and running before their competitors that's why super micro computer focuses on time to Market as their key metric of success and they've seen a lot of success this year with their stock more than doubling year-to date even after sliding alongside other tech stocks smci stock is set to split 10 for one on Monday October 1st so anyone holding smci on Friday September 27th will receive nine additional shares for each one they own in theory stock splits shouldn't matter super micro computer will have 10 times the Shar at 1/10th the price and that's that but in reality studies show that companies announcing stock splits tend to rise an average of 25% over the next year versus about 12% for the S&P 500 one big reason for that outperformance is many brokerages outside the US still don't offer fractional shares so smci share price going from $600 to 60 lowers the barrier of entry 10-fold for those investors just something to consider before saying stocks puts don't matter all right I know this video is a little long but I want it to be thorough since investors might be missing the big picture by only focusing on Nvidia stock ahead of earnings nvidia's AI chips are unlocking huge investing opportunities in a wide variety of Industries but they all rely on solid server infrastructure high bandwidth memory and the machines that build them and in my opinion the way to find the winners in these quickly growing Hardware markets is to understand the science behind the stocks that's the best way to get rich without getting lucky so research like this
Comments 0