저자: JeremySiegel · 외국인
요약 보기
금리 상승(10년물 24년만의 고점 부근)과 유가 불안으로 금융시장 긴장. Siegel은 TIPS 3.35%와 주식 대비 여유 마진 축소를 지적하며 금리 상승이 고마진 기술주 외 실물 기업 이익을 압박한다고 봄. 연준은 연내 추가 인상이 필요하다는 견해. Box CEO Aaron Levie는 AI 에이전트가 소비자 기술의 큰 기회라 주장하나 투자조언은 회피. 엔비디아는 PER 14~15배로 저평가로 제시되고, A100 GPU의 수명 연장과 AI 인프라 수요 초과(공급 부족) 전망을 언급; 반도체(AMD, 마이크론, SK하이닉스 등)의 모멘텀 확인.
다듬은 글
The decisive moment begins with some reduction in rates, as the 10-year yield retreats from 24-year highs. You get your scorecard 60 minutes before the end of the trading session. There 's a little tug-of-war going on here between the reduction of interest rates and oil, which has been stubbornly clinging to some headlines about possible additional assets in the Gulf.
And that brings us to our talk about the tape this afternoon, about how you might position your portfolio as we begin the final quarter of the year. Let's include Jeremy Siegel, professor of finance at the Wharton School and chief economist at WisdomTree. Professor , I'm glad to have you. Thank you for joining us.
I'm glad to see you, Carl. Have you noticed the beginning of a bond rally today? Well, I'm relieved we had it because I really couldn't believe the returns I was seeing. Well, well, I particularly like to look at the yields on Treasury inflation-protected bonds, and the 30-year bond reached 3.35%, and I don't think we've seen that in 20, 25, or 30 years.
Yes. And now, is that competition? You know , when you get almost 3.5% after inflation, a stock market with a P/E ratio of 20 gives you 5% after inflation. Therefore, there is still a good margin, but that margin is shrinking. And, in my opinion, one of the reasons why the technology has held up so well, I mean, look at the margins of the technology, of Mag 7, you know, they're 50%, 60%, or 70%.
Therefore, there is an increase in interest rates . But, you know, outside of technology and in the real world, when they're working with margins of 7, 8, 9 or 10 %, you start raising interest rates and you really cut into our profits. So, you know, for ex ...you know, we all thought we had this big rotation going in the first half of this year and it certainly looked like it was going to work, but I think this rise in interest rates is really, you know, I put a stop to it.
Correct. Are the bulls right to be encouraged by what Jefferson said today about taking more time to capitalize on what we got from Auto Waller? Yes, I mean, you know, obviously the war is in a really difficult situation for the next meeting. There are 6 days until the midterm elections. Honestly, I think we need two more increases this year, but I think I might be able to convey two more, but I think I could say to the committee, "Listen , guys, if you wait, maybe we'll do 50 basis points in December .
Let's not make this political by doing it six days before the midterm elections. Do you know what a difference a month or two makes at 20 or 25 basis points? So, I think he has another juggling job to do there. But I mean, when you look at the futures market and some of the others, I mean , this is the strength of the economy, and the rising yield just tells you that, you know, this economy can withstand an increase in short-term interest rates.
Yes. Yes, good article by Greg Ip today in the Journal arguing that a nominal GDP of 6-3 pulls back to a 10- year Treasury bond at 5.5 without any problem. Yes. Yes, and we normally think of the 10-year Treasury as being 80 basis points above the federal funds rate. If You're at 5, 5.25, or 5.5, that's way above where we are right now relative to the normal slope.
Some people have said that the 2-year bond is a good indicator of where federal funds should be, and boy, is that true in all those temporary structure issues, too. Meta just closed its best month since November 2022, the month after ChatGPT first launched, as its new AI assistant, Muse, has taken the streets and the world by storm .
According to Sensor Tower, Muse has surpassed 5 million downloads since its launch just 22 days ago. Our next guest says that AI agents will ultimately be the biggest opportunity and revolution in consumer technology since the App Store itself, and his company just announced a partnership with Muse.
We're going to introduce Aaron Levie, co-founder and CEO of Box. It's great to see you again, Aaron, it's been a while, and... Yes, thank you. I think, to be honest, when Claude and all that They were going swimmingly, it took me a while to understand what those tools were. Now I get it . I think we all do.
How much more can Muse grow? Well, as you said, I think those are relatively small numbers compared to the total market this kind of technology is targeting. And obviously, Meta has a fantastic distribution channel for Muse to launch; now through ChatGPT with dots, another great entry in the sector, and they have a solid distribution channel.
There are startups like Instinct, and I'm sure we'll see something from the Claude and Google universes. So this is shaping up to be a clearly very dynamic category and space, but I think there will be room for some different winners, just as we've seen in areas like mobile operating systems and even the AI platforms themselves.
Is it too early if this is like the launch of the App Store? It took years before Uber and Airbnb appeared on the scene. Is it still too early to know how to invest in this, or are the winners , like Meta and others like it, already staring us in the face? Well, I don't really think I can offer investment advice on this front.
But I do think you can extrapolate, you know, maybe we'll reach a point where there are a few hundred million users of these kinds of personal agents; and what kind of economic activity will a user or consumer provide to their agent, and how will that activity be executed? One of the biggest changes is that in a world of agent task execution, both the opportunity and the risk is that the end user's preference starts to become less important than what the agent is able to execute.
I might have a preference. Traditionally, I prefer where to buy something or where to carry out the transaction. But if an agent can find a better price or a better opportunity for me, I might end up going along with what that agent decides or recommends. So, that has the opportunity to provide a small market shift in some of the dynamics based on who is a holder now versus who will be one later.
But unfortunately, that doesn't tell you exactly what to invest in, because you also have a huge opportunity for holders to respond in very different ways. Effective. There's a kind of increasingly famous phrase in Silicon Valley that goes: "Build something that agents want." And this is based on the kind of startup spirit of creating something that people want, which emerged, I think, from Paul Graham and Y Combinator many years ago.
And the idea was, you know, to create a tool that people would want to use, that would make them happy or successful, or save them money or help them be more efficient. Now, what we're seeing is that startups are doing the same thing, but for agents. And that could be the infrastructure in a company.
Create the infrastructure that agents want to work with or create the APIs that agents want to be able to use to perform online transactions. And therefore, I believe that what will increasingly happen is that the companies that are best prepared to work with agents will be the ones that drive the next cycle of growth in consumption or even in companies.
If we look at Nvidia today and compare it with som very interesting. Nvidia trades at around 14 or 15 times its earnings, while Apple is closer to 35 times. Now, add a growth difference. Nvidia is expected to grow by around 17% year-on-year. And the problem is not a lack of demand, but quite the necessary .
Therefore, we , deeply integrated into the AI ecosystem, and that continues to trade at valuations that some investors consider surprisingly low. For some analysts, that makes Nvidia a much clearer opportunity than other, slower-growing technology companies . But there is another part of Nvidia's story that doesn't receive enough attention.
How long will your AI infrastructure actually remain useful ? For a month, investors were extremely quickly. A new architecture arrives, companies buy it, and the previous generation depreciates. And in a few years, the old hardware supposedly becomes obsolete. But that assumption may be too simplistic.
A cloud infrastructure operator recently highlighted an important example related to Nvidia A100 GPUs. Thoat infrastructure has been re-contracted to get more workload until 2029. And that's a all AI workloads require the new GPUs. The latest hardware may be ideal for training the most advanced fundamental models , but the older accelerator can still be useful for AI physical batch computing , .
Therefore, when the new generation arrives, the older hardware does not necessarily disappear. You can simply switch to a different workload. And that means the lifespan of AI infrastructure could be much longer than some investors originally expected. And if a GPU acquired years ago can continue to generate revenue for several , take a look at Nvidia's latest infrastructure release.
The first system improvements have already begun to reach customers. Some early implementations involve setting up NVR 72, but obtaining the GPUs is the only part of the process. A functioning AI cluster needs GPUs, networks, data center capacity, power, and software. Everything has to work together.
You can't deliver 95% of the infrastructure and consider it finished. The system must work, and that' so important. They convert individual pieces of hardware into usable computing capacity for businesses and AI companies. Right now, demand is exceeding supply. Some suppliers claim that virtually all the GPUs .
And that's how useless this individual AI has become . Of course, the market eventually responds to scarcity. More data centers are being built and more power is being connected. More GPUs are being manufactured and more capacity is being deployed in the cloud. Over time, supply and demand come closer together.
The question is not whether the imbalance will last forever. He won't. The real question is how long demand can continue to outpace supply. And then we have long- term perspectives that suggest the imbalance could remain significant for years. At the same time, money is circulating in the semiconductor industry.
AMD, Micron, SK Hynix, Marvell, and several AI-related semiconductor stocks have been gaining momentum together. And that 's important because market leadership is constantly changing. When a group becomes full and momentum slows, capital moves elsewhere. And 2026 has increasingly become a story of these rotations.
The rise in crude oil and Treasury bond prices could put pressure on areas such as health, finance, and small businesses. However, AI-related technology stocks can regain strength when earnings and expectations remain solid. Therefore, investors don't just wonder if the market will go up or down. They ask themselves : "Where will the money go now?"
Comments 0