저자: Asymmetric Investing by Travis Hoium · 외국인
요약 보기
이 영상은 최근 48시간 동안 시장이 크게 흔들린 배경과 오늘의 급락 원인을 짚는다. 먼저 히트맵을 통해 반도체와 반도체 장비 업종의 약세를 보여준다. 데이터센터 투자가 줄면 칩 수요가 감소하고, 남은 기업들이 더 치열하게 경쟁하면서 매출 기대치와 마진이 낮아질 수 있다는 논리다. 예시로 내년 칩 수요가 50% 성장 기대에서 20%로 낮아지는 상황과, 알파벳이 자체 TPU를 늘리면 엔비디아 칩 구매가 줄 수 있다는 가능성을 든다. 또한 AI 지출이 실제로 얼마나 이뤄지는지에 대한 의문이 제기된다. 코딩 외 수요가 충분한지, 애플의 일일 지출 한도 하향 소식, 에이전트 기반 서비스의 저조한 실적(우버 이츠 약 500건 주문 중 라이드 4건) 등이 언급된다. 마이크론의 대만 공장 건설 중단 소식도 AI 인프라 병목 관점에서 경고 신호로 소개된다. 업종별로는 통신, 공공요금 관련 소비재, 디즈니, 헬스케어, 보험, 필수소비재, 에너지가 상대적으로 강세를 보였다고 설명한다. 엔비디아 사례에서는 현재 낮아 보이는 PER(약 23배)이 과거 대비 비정상적으로 높은 마진(매출총이익률 약 75%, 영업이익률 약 65%)에 기반한다는 점을 짚는다. 영업이익률이 20% 수준으로 회귀하면 실질 멀티플이 크게 높아질 수 있다는 분석이다. 마지막으로 발표자는 사전 리서치를 통해 비대칭 포트폴리오를 운영하며, 지수가 20% 또는 30% 하락할 때 기계적으로 추가 매수하는 원칙을 밝히고, 감정을 배제한 투자의 중요성을 강조한다.
다듬은 글
Today the market is volatile. As an investor, you probably want to know what's going on. Over the last 48 hours, everything seems to have flip-flopped several times. There was an election, and when the news started coming in Tuesday night, stocks fell. Then Wednesday was a very good day.
And today, Thursday, stocks are down quite a lot. I want to look at where the market is weak, what's driving this, and how to think about it going forward. At the end of the video, I'll also talk about what I'm doing as an investor today, whether I'm buying to take advantage of the opportunity. A little, yes. But I've built my portfolio so I can move aggressively when the market is at its weakest. I don't think this selloff is over, and I'll explain why.
Semiconductors and semiconductor equipment stand out here. The colors are all over the place, and the market's thinking goes like this: if fewer data centers get built, total demand for chips falls. When demand falls, the remaining companies have to fight harder over that revenue. That means revenue could come in below expectations, and margins could be somewhat lower too.
For example, if next year's chip demand was expected to grow 50% but only grows 20%, companies' revenue shrinks by that much. Competition also intensifies. If Alphabet makes more of its own TPUs, it buys fewer Nvidia chips. That could be a reason Nvidia is falling.
What the market is looking at right now is whether a real slowdown is coming. You could say catalysts for a slowdown in capital expenditure (capex) are starting to show up. Companies that had been tailwinds are turning into headwinds. The margin compression I mentioned could appear, and growth rates could come down and slow. What we're seeing today is multiple compression. Even though Palantir put up strong results earlier this week, that doesn't mean its results have caught up to the multiple the market currently assigns it. That's what the market is pricing in right now.
Another steady source of pressure is the question of how much money is actually being spent on AI. I've been talking about this for a while. AI is amazing for coding, but who else is using it? Are ordinary people using AI or using it for shopping? Those parts have started to wobble. Companies that were spending the most on AI have actually started cutting back. In today's news, Apple lowered its daily spending limit from $300 to $200, and there are expectations it could go down to $100. AI spending is still large. But if you look at revenue at companies like Anthropic or OpenAI, only a tiny number of people spend millions of dollars on tokens, and most of that is for coding.
Over the past few weeks, the bull case for AI agents has also started to unravel a bit. First, Muse isn't as popular in the App Store as it was. People seem to try it and leave relatively quickly. I'm one of those people. I disconnected everything I had hooked up to Muse. It didn't adequately serve a purpose that was valuable to me. There wasn't the 10x improvement that would actually change how people operate. And there's data too. The thing getting the most attention this week is Uber. There were roughly 500 orders that came through AI agents on Uber Eats. Only four were rides. Four completed rides through agents, total. I think one of those was Brad Freeman, whom some of you may follow on Twitter, and he said it was a much worse experience than using the app directly.
On the demand side, the question to ask is this: is there demand beyond coding? Are the business models that were expected to drive the future of AI and infrastructure buildout breaking down? Is this a house of cards? Another piece of news today is that Micron halted a $4.5 billion factory construction project in Taiwan because of a labor dispute. I'm not sure how that will play out. But Micron sits at the center of one of the biggest bottlenecks in AI infrastructure buildout. If they might have to pay workers more and they say they won't just push through, that could be another warning sign for this industry.
So there are a lot of questions, and semiconductor and semiconductor equipment companies are falling together. Let's look at related companies. Communication services, in other words advertising companies like Alphabet and Meta, haven't dropped as much, but they're down today. On the other hand, there are some areas showing a bit of strength in the market: the daily bills people pay for phone and utilities. T-Mobile, Verizon, and Comcast are doing well, and Disney looks pretty good today too. Tesla may be a bit different. Tesla was a big part of the AI infrastructure buildout. CEO Elon Musk's political involvement is also significant, so it could face political pressure. It's not just Tesla. Will SpaceX's contracts with the government face more scrutiny going forward? That's entirely possible. If Tesla and SpaceX merge, one of the reasons investors bought Tesla over the past few months could become harder to sustain.
Still, you should look at the areas that are strong. If you're down a lot today, you probably don't have exposure to sectors like healthcare. Healthcare is having a very good day today, because even if a recession hits and AI infrastructure spending stops, people will keep buying products from Eli Lilly or Johnson & Johnson. Financials are a bit tougher, but insurance rose quite a bit today. What's caught my eye recently is consumer staples. A lot of stocks in that area have become relatively cheaper. Not all of them. Walmart and Costco have been fairly expensive. Still, these are the kinds of companies that can hold up even if there's a mild recession. The best-case scenario for them is that AI turns out to be less disruptive than many investors expected. Even if the market corrects, it's good if people can keep their jobs and spend money on things like oil and gas. Oil and gas rose again today.
One more thing: it depends on which market and which index you're looking at. Until now I've focused on the S&P 500. This is a Nasdaq 100 heat map. Almost everything is red, mainly because these are tech-focused companies. Overall, it's a very rough day. SpaceX is down 10%. But the consumer staples companies included here are doing fairly well. Let's also look at the Dow Jones Industrial Average. What's surprising is how big a role tech companies play in the Dow too. Other industrials, healthcare, and consumer staples are relatively okay.
Let's use Nvidia as an example, because it's one of the largest companies in the world and it's one where the pressures investors are worrying about today could materialize over the next few years. You know Nvidia has put up excellent results over the past few years. Look at its growth rate: 83% over the last year. Investors often hear that Nvidia is cheap because its P/E is very low. But if you break down the factors supporting that P/E one by one, the story changes. This applies to tech in general, and to today's AI-related trades as well: a stock that looks cheap isn't necessarily cheap. That's because this is based on trailing numbers, not forward ones. On a trailing basis, Nvidia's stock looks cheap, especially for a company growing this fast. But you have to look at it in historical context. Nvidia was not always a steadily high-growth company. In fact, its growth over the last few years is something we haven't seen in 15 years. So I pulled up data going back to 2010. Nvidia has periodically had negative revenue growth. What happens if revenue growth turns negative in 2028? The stock's multiple will likely compress. Price pressure is also likely.
Now let's look at margins. This is the part investors are most worried about today. A P/E of 23 assumes margins that Nvidia has never had before. Gross margin today is about 75%, and operating margin is about 65%. Historically, going back to 2005, gross margins were mostly in the 50s. They rose to about 60% in 2022. What happens if margins fall by 15 percentage points? More important is operating margin. Nvidia's operating margin has historically gone negative at times in its cycles, but it was typically in the 15–20% range. If operating margin drops from 65% to 20%, a P/E that looked very cheap at 23 suddenly becomes roughly 65–70 times. Then it looks very expensive. At the same time, you'd have a stock with slowing growth and margins under pressure. The market has already started pricing in some of these things today.
시장은 미래를 먼저 반영하는 메커니즘입니다. 그래서 AI 인프라 투자가 조금이라도 둔화되면, 성장률이 마이너스가 되지 않더라도 이미 주가에 반영된 수준의 성장이 나오지 않는 것만으로도 문제가 될 수 있습니다. 모두가 공급을 늘리고 있고, 새로운 생산능력을 만들고 있으며, 제품을 만들고 더 효율적으로 만들고 있으니까요.
지금 건설되고 있는 것들을 감안하면 2027년에는 토큰 공급이 과잉될 가능성이 충분히 있습니다. 오늘 시장이 걱정하는 것이 바로 이것입니다. 수요 측면이든 공급 측면이든 어느 쪽이든 둔화가 생긴다면, 그 이유는 채권시장이 이 투자를 자금으로 뒷받침하지 않을 수도 있기 때문입니다. 주식시장 역시 이를 자금으로 뒷받침하지 않을 것입니다. 그리고 주가가 떨어지면 이런 프로젝트들의 자금을 조달하기도 더 어려워집니다. 그래서 지금 AI 투자 전반이 하락 나선형에 빠진 것처럼 보입니다. 이것이 오늘 시장이 생각하고 있는 것입니다.
이제 제가 지금 무엇을 하고 있는지, 그리고 요즘 투자자로서 어떻게 생각하는지 말씀드리겠습니다. 저는 리서치를 미리 해둡니다. 오늘 당장 현장에서 리서치를 하고 오늘 결정을 내리려 하지 않습니다. 내일이나 모레 무슨 일이 일어날지를 맞추려고도 하지 않습니다. 비대칭 투자는 향후 10년에서 20년 동안 시장 대비 내가 가진 우위에 관한 것입니다. 그래서 스포트라이트 기사를 쓸 때는 해당 기업을 깊이 파고들어, 향후 10년에서 20년 동안 이 기업이 10배 성장할 수 있는 투자 논리가 어디에 있는지 설명합니다. 이렇게 하면 이런 날에 한 순간의 상황에서 벗어나 조금 더 명확하게 생각할 수 있습니다.
비대칭 포트폴리오에는 제가 사고 싶은 기업들이 들어 있습니다. 그중 하나가 10% 떨어졌다고 해도, 지난 24시간 동안 실제로 달라진 것은 거의 없습니다. 오히려 흥미로운 매수 기회일 수 있습니다. 또 하나 제가 해온 방식이 있습니다. 비대칭 포트폴리오를 예로 들지만, 개인 투자에도 같은 방식을 적용합니다. 주가가 떨어질 때 더 많이 사는 것입니다.
비대칭 포트폴리오에는 트리거를 설정해 두었습니다. S&P 500이나 나스닥 100이 20% 떨어지면 투자 금액을 두 배로 늘립니다. 월 500달러에서 1,000달러로 올리는 식입니다. 30% 하락하면 500달러를 추가로 더 투자합니다. 남들이 두려워할 때 나는 탐욕을 부리는, 제가 실천하고 있는 방식이 바로 이것입니다.
저는 매일매일 시장 타이밍을 맞추려고 하지 않습니다. 내일이나 다음 주에 주식이 오를지 알지 못합니다. 하지만 시간이 지나면 주식시장은 오릅니다. 그래서 미리 리서치를 해두고, 기회가 올 때마다 기계적으로 매수한다면 이것이 이 상황을 생각하는 가장 좋은 방법이라고 생각합니다. 이렇게 하면 투자에서 감정이 빠집니다. 그리고 그것이 우리 인간이 투자자로서 가장 어려워하는 일입니다. 지금 느끼는 감정은 앞으로 며칠 동안 시장에 대해 느낄 감정과 다를 가능성이 큽니다. 그리고 많은 경우 가장 좋은 답은 그냥 아무것도 하지 않는 것입니다.
투자에 대해 더 많은 제 생각이 궁금하시다면, 주말에 많은 콘텐츠가 올라갈 예정입니다. 제 뉴스레터인 Asymmetric Investing을 확인해 보세요. 화면에 링크가 있고, 설명란에도 링크가 있습니다. 유튜브에서도 더 많은 영상을 올리겠습니다. 구독해 두시고 제 콘텐츠를 모두 받아보세요. 시청해 주셔서 감사합니다. 다음에 또 뵙겠습니다.
댓글 0