원문 보기
Over the past few weeks, there's been a clear pullback in AI-related stocks, and that question in many people's minds has suddenly popped up again: has it peaked, or is this just a halftime break? Should one take advantage of the rebound to run, or should they get back on board when others are fearful?
It's at a time like this that Tom Lee has stepped forward once again with a very clear stance: he doesn't believe this is a major top; instead, he feels it's more like a healthy consolidation, and the market still has a chance to move higher, or even reach new highs. In other words, his point is quite direct: those who sell AI stocks out of fear now will likely regret it later.
Let's take a look together. Macro update and top ideas webinar. It's very timely because as you know, there's been painful corrections taking place in parts of the market that have been very hot. And I think the question in everyone's mind is of course, is this a topping process leading the market lower or are we in a healthy consolidation and we continue the uptrend?
I'm in the camp and I believe in...news this month in Tom, there seems to be...and this was addressed before, there seems to be a lot of concern about increasing memory prices ...impacting earnings and squeezing margins of AI related companies. I know you're bullish mid to long term...I mean, it's very messy.
Whoever asked that question, because for somebody who thinks this is a bubble and companies are just spending money without thinking about it...it's a...it's...a very, very shallow take. The Mag 7 companies...companies...understanding ...so much money hyper, it's not because they think stock...are you confident that home ownership...is going to come back in line with the next generation of homeowners, and if so, why?
Yeah, that's a good question. I think it's a fact pattern that people have delayed buying houses, but...it's understandable. I mean, housing is not cheap. I mean, even rents are incredibly expensive. I think inflation cumulatively has had a negative effect on young people. But the same people who think no one's going to buy a house are the same people that told me in 2009-2010 things that they were absolutely certain about, and these were economists...big brain people, and the ones everyone listens to.
Number one, they said no one was going to buy houses in 2009, and actually that was the bottom. But secondly, they said nobody was going to drive cars. They were saying young people weren't even going to get driver's licenses, and so car sales were going to actually be stuck at that was 10 million below peak .
As you know, that didn't play out. In fact, plenty of young people are actually buying cars, electric cars. So you know, and maybe more recently, I remember in 2018-2019 a lot of people told me that the trend was small homes. No one was going to buy big homes again . So they were abandoning properties in like Fairfield County and saying that the only thing people wanted were tiny footprint homes.
You know what? 7 years later, do you know how many people regret buying a tiny home when they actually wish they bought land? So I'm going to say...nothing has changed for centuries. People bought homes and lived in, and land has been a great source of future value growth and wealth creation. I don't think anything changed.
Are the Mag 7's hundreds of billions really just burning money? What is the market most worried about now? They are worried about rising memory prices in the AI chain, especially High Bandwidth Memory, or HBM, which is essentially a key component on GPUs. You can understand it as a crucial link in the training and inference system.
HBM price hikes mean higher server costs and greater data center investment, so will corporate profits be squeezed? Will AI companies be unable to make money? These concerns aren't baseless, so when the market gets nervous, AI stocks tend to fall first. But Tom Lee's response is very direct. He says that simply understanding this as bubbles and companies wasting money is a very shallow perspective.
Because the ones really spending big money aren't edge cases or small companies, but "Mag 7" level companies—the super platforms like Microsoft, Google, Amazon, Meta, Apple, Nvidia, and Tesla. They aren't startup companies that have never seen a cycle, nor are they companies looking to tell a story for the next round of funding.
They are among the companies globally that best calculate return on capital. To put it plainly, these companies wouldn't drop tens of billions in annual expenditures just because they think spending money looks cool. Many people's understanding of AI investment remains at a very superficial level: companies spend money to buy GPUs, the market gets excited, the stock price rises, and once everyone discovers that profits aren't keeping up, the stock price collapses.
But the deeper logic is that super-platforms are frantically expanding computing power not to signal to the market, but because they clearly see that whoever secures computing power, models, data centers, power, software stacks, and ecosystems first will have the best chance to capture a larger profit pool in the coming years.
This is not about sentiment; it is about industrial logic. You can look at Microsoft, right? Why are they willing to continuously push capital expenditure higher? Because AI is not a standalone product; it will redefine the value of cloud services. In the past, customers bought cloud for servers and storage; now, they are buying intelligent capabilities, model access, AI agents, enterprise knowledge bases, and automated workflows.
Once these things are running, customer stickiness is much stronger than traditional cloud. Revenue per customer is also likely to be higher. Amazon is actually the same; AWS isn't just selling machines, it's selling the entire digital foundation. Google goes without saying; search ads, YouTube, and cloud are all being rewritten by AI .
Meta putting AI into ad recommendations and content distribution provides a small boost in effectiveness, but acts as a huge lever for profit. So, what I really oppose isn't market concern, but rather lazy-minded concern. You cannot see capital expenditure soar and automatically jump to the word "bubble.
" Capital expenditure itself is not a sin; the key is who is spending it, why they are spending it, and whether they can build a deep moat after spending it . How could an industry with blown-out supply be driven by fake demand? That is why I think this Tom Lee report is more valuable than just shouting that AI will keep rising.
It is actually a reminder not to use the lens of old-era cyclical stocks to view this round of AI infrastructure reconstruction. This AI wave is not like many concept-driven hypes; it has a hard reality behind it, where real demand is already happening, transmitting from the base layer upward . First, there weren't enough chips, then HBM became tight, and then switches, optical modules, power, liquid cooling, data centers, and transformers all became bottlenecks.
Once you see this phenomenon, you should realize this is not an industry that tells stories without orders; this is an industry where the upstream is already exploding with demand. The market is currently anxious about short-term pressure on profit margins, but the industrial reality is that supply cannot keep up with demand.
Of course, it must be added here that tight supply does not mean all stocks are worth buying into. Being bullish on AI does not mean that just any company claiming to do AI can take off. The true beneficiaries will most likely still be those few categories. The first category includes super platforms that control traffic gateways and enterprise clients.
The second category consists of "shovel sellers," such as those in chips, memory, networking, cooling, and power infrastructure. The third category includes software companies that directly translate AI into revenue and efficiency gains. As for those companies that only repeatedly claim to fully embrace AI at press conferences, but show no orders or improved cash flow in their financial reports, the risks remain very high.
The pitfall many retail investors fall into is equating the AI mega-trend with the idea that all AI stocks will rise; these two are actually not the same thing at all. The internet was a mega-trend back then, but not all internet companies profited . New energy was also a mega-trend, but not every company in the sector thrived .
True market opportunities often lie in companies with pricing power, cash flow, and execution capabilities within the industry, rather than the companies that make the most noise. Looking at it from a business perspective, why might those selling AI stocks now truly end up regretting it? Because the market is best at creating an illusion: whenever a correction occurs, everyone assumes the logic is broken.
But often, it's not the logic that falls; it's the pace , the positioning, and the overly inflated short-term expectations. Especially during a strong main market trend, the truly terrifying thing isn't the volatility, but the fact that you might throw away your core assets during that volatility. Look back at many major historical trends, and they all followed this path.
In the early days of cloud computing, the market also doubted the massive capital expenditure and questioned when the profits would materialize. Before the smartphone supply chain exploded, some thought phones were merely cyclical consumer electronics. Even e-commerce, media, and data centers—things we take for granted now—all faced widespread skepticism in the early days , such as "too expensive,""overvalued," or "just telling stories." The point is that once an industry truly enters the infrastructure deployment phase, stock prices may not rise linearly, but the trend is often much more enduring than most people think.
So, what is the biggest difference between this wave of AI and the internet era of 2000? Many people love to bring up 1999 and 2000 to scare others. This comparison isn't entirely baseless, but it often misses the point. The biggest issue in 2000 was that many companies lacked real cash flow, relying instead on capital markets to fund expansion, so demand couldn't keep up with supply.
This current wave of AI investment isn't driven by small concept-driven startups , but by tech giants with the strongest cash flows and solid balance sheets globally. Much of their AI investment can be covered by their current business profits, which makes a huge difference. This doesn't mean prices won't drop or valuations won't correct; short-term volatility is inevitable, especially when markets worry about interest rates, earnings realization, or regulatory hurdles.
Tom himself isn't saying it will go straight up; he suggests these current fluctuations look more like healthy consolidation rather than the end of a major trend. You’ll find that this approach aligns more with mature investing—it doesn't deny risk, but rather distinguishes between structural risks and emotional ones.
After hearing these points, the real question isn't whether AI stocks are still worth buying, but whether a company is a true beneficiary of AI capital expenditure and can convert that investment into revenue, profit, or a moat. How much optimistic expectation is already priced into its valuation? If it drops another 10%to 15 %in the short term, would you sell because the logic changed, or just because seeing red numbers is painful?
The most important takeaway from Tom Lee this time isn't that he's bullish again, but that he's dismantling the controversy around this AI wave. He uses a model that exposes typical market misjudgments. People always have a habit of mistaking current discomfort for the end of a growth cycle; housing is expensive?
Then they say no one will buy homes in the future; young people are saving money? Then they say no one will buy cars; AI capital spending is too aggressive? Then they say it must be the peak of a bubble. History has proven time and again that major trends rarely end when people fear rising costs, but rather when they no longer see new value, new demand, or new returns—and the AI market in July 2024 is clearly not in that state.
On the contrary, we see demand expanding, capital increasing, infrastructure gaps being filled, and industry leaders securing their positions. The market isn't worried that nobody wants AI, but rather that so many want it that costs might spike first. These two trends may both appear to drive stock prices down, but they are fundamentally different things.
The former is a story going bankrupt, while the latter is merely the growing pains of development. So, what exactly is this round of pullbacks ? From Tom Lee's perspective, it looks more like a round of filtering. He is not denying AI, but answering who the true beneficiaries are. Who is just riding the trend?
Who can turn their capabilities into a moat? Who is just burning shareholders 'money? Who will hit new highs after the volatility? Who will have their true colors revealed in the next round of earnings? Ultimately, Tom Lee has never just been bullish on a few hot tickers or a rebound in market sentiment; he is bullish on something much bigger: that AI is moving from sounding impressive to actually changing revenue structures, profit structures, and the competitive landscape.
Once you realize this, you will understand why those rushing to sell now will most likely regret not just missing out on a few points, but selling themselves out halfway through a massive structural shift in the industry. For more videos, click the bell below to join the second channel. See you next time.
Comments 0