저자: TomLee · 외국인
요약 보기
결론: 현재 AI 관련 매도는 막바지이며 조정은 건강한 과정이고, 저가 매수 기회라는 낙관론. AI 버블 논쟁: AI 관련 우려가 버블을 막고 있으며, AI 발견(수학 문제 770개 해결 등)이 계속되는 한 투자 수익이 발생한다고 주장. 부채 시장은 AI 과열을 완충하는 역할을 한다고 평가. 투자 규모: AI에 향후 25년간 30조 달러가 투입되어도 미국 가계 이자소득(연 약 7조 달러)으로 조달 가능하다고 주장. 모바일 네트워크·고속도로 각 약 25조 달러 대비 AI가 더 중요한 인프라라고 평가. 매수 타이밍: 기술주 매도가 컨센서스가 되어 최종 단계라고 판단, 매수에 긍정적. 금리는 5%에서 하락 중이며 Fed가 매파에서 비둘기로 전환 가능성을 언급. 남은 불확실성은 신용시장 경색과 전쟁. 매수 대상: AI, 메모리 병목, 엔비디아, Mag 7이 과매도 상태로 바닥권이라고 봄. 소프트웨어와 크립토(비트코인)에도 긍정적. 광범위한 경기민감주와 금융주(금리 하락 수혜)도 기회로 제시. 시장 해석: 반도체 지수 및 DRAM ETF의 급락을 보며 현 상황을 '압축된 스프링'으로 표현, 펀더멘털은 안정적이라고 주장. 역사적 비유: 1997년 인터넷 초기 시스코가 Greenspan의 '비이성적 과열' 발언 이후 50% 급락했으나 이후 회복했던 사례를 제시하며 현 상황이 유사하다고 주장. ChatGPT 출시 3주년을 맞아 AI 여정은 초기 단계라고 평가. 리스크: 신용 시장 폐쇄 상태와 전쟁에 따른 불확실성이 남아 있다고 언급. 기타: 후반부는 Nvidia 유럽 행사(AI 팩토리, DGX Pod, 에이전트 시스템 소개) 및 로봇 홍보 장면으로 투자 관련 발언 없음.
다듬은 글
You say that fears about AI are preventing it from becoming a “ bubble.” Some would suggest that these concerns are only causing an already formed " bubble" to burst. God, I know there are going to be two sides to this debate , but I argue that the fears about AI are preventing a " bubble" because we continue to see amazing discoveries made with AI.
I mean, whether it's Muse and agents, or maybe a few weeks ago, OpenAI, which solved 770 math problems. This is not the end or late stage of the cycle for AI discoveries and even for productivity growth. So, as long as we see progress there, every dollar of capital raised for AI is generating returns.
I think it's actually very useful for debt markets to act as a kind of buffer against the uncontrolled speculative boom in AI. So I think this is actually a very healthy correction. So you're focused more on the spin- offs, you might say, than on the issues of cost, funding, capital shortages, or anything like that?
Yes, unfortunately, AI really does require a lot of money. A few weeks ago, we did a report that shows that if AI spends $30 trillion over the next 25 years, then, first of all, there are plenty of investment dollars, because the interest income of US households alone is about $7 trillion per year. So, you know, a trillion of that amount can be financed just by the interest income of Americans.
But building a mobile network cost about 25 trillion, and highways cost 25 trillion. I think AI is probably a more important infrastructure development than mobile communications, and it costs about the same. So I don't think there's too much money being poured into AI today . Would you buy during the sale?
Yes. I think we are in the final stages of the sell-off because the play against the technology sector has become the consensus. And I think that bond yields, which were very worrying for the markets, as you know, have receded . So you have two factors that used to be obstacles: the Fed was quite hawkish and now it's potentially becoming dovish, and interest rates, which were approaching 5%, and people were afraid of 6%, are now back below 5% .
I think the only thing left is that the debt markets are somewhat closed at the moment , and of course, the uncertainty about the war. That is, if you say "yes, I would buy it," then what would you buy? Would you be left with only the biggest AI-related stocks on the market? Yes, Scott, I think AI, memory bottlenecks, Nvidia, and Mag 7 -- they're selling off because it's easy to play short right now , and of course people are worried about the credit aspect.
I think they are hitting rock bottom , but I also like the software and I am leaning towards crypto. I think I'm pretty optimistic about broad cyclical trading. And, as you know, the financial sector is not growing, but rates are falling. So, I think there are a lot of possibilities. I view this as a compressed spring, as the fundamentals remain stable.
Wow, I mean, if you look at, for example, the semiconductor index or the DRAM ETF, we show that. How much destruction in this area. I wonder what story this graph will tell in the coming weeks, if not months. Scott, I...I think it just shows what a struggle it is today around AI. But remember, in 1997, 3 years after the launch of the public internet and browsers, Cisco, which was the biggest stock at the time, had a huge 50% drop, partly because Alan Greenspan talked about " irrational optimism" and later foresaw a preemptive rate hike, very similar to today.
Cisco fell by 50%, dropped to five, and a year later it was 18. I think the situation is similar. You know, ChatGPT is actually approaching its third anniversary. I think we are still at the very beginning of the AI journey, but people's excitement is understandable. Many people have made money, and profits are being fixed.
Hello, guys. Oh, really? Okay, let's go. Сюди, підійдіть сюди. Show people that you can walk. І у вас є камера. Oh, right? Не спіткніться. Can you sit down? Okay, get up. Sleep. Лягайте спати. Вам потрібно відпочити. Добре, вставайте. Did you know that you were created here? You were created on RTX Spark.
I know. You were trained in the cloud on a bunch of Nvidia supercomputers, and you learned your skills right here on the RTX Spark, in a digital twin, the digital twin of Omniverse. Хіба це не неймовірно? I know. I know. You are so cute. And your parents' names are Pollen. P O L L E N. Чи не так? Aha.
Pollen Robotics. And, apparently, everyone likes you so much that everything is sold out. I know. So if you go to the Pollen website, everyone who watches it knows, and 20 million people will watch us. Everyone who watches it, go to the Pollen website and place an order. I think there are 10 people in this startup company.
They could become stars. Це буде так захопливо . Кожен має такий мати . Ви такі милі. So, we'll have RTX Spark, and we'll teach you new skills. Isn't that right? Oh, oh. Oh, oh. Oh, look at this. This is incredible. Ти вмієш робити сальто? This is so cool. This is so cool. Wow. Good. Ну що, вгадайте що?
So, what did we talk about today? We talked about how Europe is building its AI economy. And it all starts with infrastructure. We announced a brand new system called DGX Pod. It facilitates the creation of AI factories. And these AI factories are no longer just running models. WARNING. Focus. Come here.
Come here. Come here . Focus. And these AI factories don't just launch...That's how it is. Sit down. Yes. They...They're not just running AI models anymore. They run entire agent systems. And these agents are made up of five things. There is a large language model that is responsible for perception, reasoning, planning, and action.
And it works together with a shell, which essentially turns it into an exoskeleton. This shell runs inside a secure runtime environment that we call Open Shell, which has been supported by so many people, and you give it access to knowledge and tools. This agent is powered by these computers, but we're also moving into the next generation , where you come in— physical AI.
That's all . This is what we talked about. Anyone can create agents. Every company in Europe will become an AI company. Without a doubt, everyone will become an AI company. And we will have a thriving AI economy here in Europe. So after 33 years—what an incredible journey, and we made a really cool video.
Want to watch it together? Okay, let's go. Come here. Just come here a little more. We will step aside. Come here . Maximum speed. Maximum speed.
Comments 0