원문 보기
with an exit multiplier of 30 and a cost of money of about 6%. Unrealistic assumptions and the fact that these were actual buyers of Cisco equipment, right? They were real customers. It was a farce. So if someone asks, "Is today's situation such a story?" This is not true. I mean, the number of people using AI is still quite small.
He is already showing himself to be quite productive. The companies that buy this equipment, hyperscalers, are very, very serious companies. They are not hyper people trying to dig up land and sell land use rights (IRUs). These are companies that actually buy, install racks, and receive large orders.
So I think we're not in a bubble yet. ongoing market headwinds, including the Federal Reserve's hawkish policy, higher oil prices, and stagnant legislative processes in Washington, cryptocurrencies like Bitcoin and Ethereum are demonstrating unexpected resilience, with Ethereum even strengthening its position relative to Bitcoin in conditions historically unfavorable for risk assets.
Financial analyst Tom Lee suggests that this surprising strength signals a major turning point in the market, especially as major financial institutions move forward with tokenized assets and stablecoin-based payment channels ahead of potentially defining changes in U.S. regulation and expanding global frameworks.
Beneath these immediate market dynamics, Raul Pal argues that an even larger structural shift is underway as society enters an exponential era where rapid advances in artificial intelligence and computing power are outpacing traditional governance and institutions. Pal predicts that the current period of volatility actually marks the construction of a new economic architecture, which could culminate in an economic singularity by 2030, where autonomous AI agents will drive mainstream economic activity and digital finance will undergo a fundamental transformation.
You know, Korea has been showing very good results not only in 2026, but also in the last few years. And the main reason is that the world is using significantly more semiconductors and memory per unit of GDP, which means that Korea as an economy and as a stock market is significantly more important in the next decade than it has been in the last 30 or even 50 years.
So profits should be at a high level. Stock markets are now so-called pricing mechanisms, and it was logical that the Korean market should have grown, but in recent quarters credit products have been added, which increases volatility. So I think to some extent this sharp move up and then the recent record drop is a leverage adjustment, because as you know, when markets go straight up, they lure in the so-called "bulls," especially those who use leverage, and then they're forced to sell.
So I think there is a forced reduction in credit positions happening now. But that doesn't mean the main story is over. Therefore, I believe that this pullback will be one of the best opportunities to buy semiconductors and AI stocks, and therefore the stock market in general. I don't know when the correction will end, but when it does, I think the current market, AI stocks, memory and semiconductor manufacturers will reach significantly higher all-time highs.
First, trying to guess the time to enter the market is never profitable. So if someone is holding these assets, they should continue to hold them. If they sold everything and got out, and are now waiting for a signal to get in, they will end up just catching up with the rising market. No one ever guesses the bottom for sure.
However, what we want to see— massive deleveraging—has already happened. If you look at data from US investment bank prime brokers, hedge funds have reduced their positions, meaning they have already sold their long positions in the tech sector at the fastest pace in almost 3 years. In fact, if you look deeper, it was the fastest pace in almost 10 years.
So we have already seen a massive reduction in leverage. And we are already seeing very high- profile stories that mass shoulder reduction has also occurred in Korea. So if people were forced to sell, they are called "weak- handed." They have already been shaken out of the market. I would say we should be pretty close to the bottom.
But again , I think the mistake people make is that they chase the market. They try to guess the peak and they try to guess the bottom. And most of those who achieved the best results were those who stayed invested. In fact, as Peter Lynch famously said: selling winners is like cutting flowers and watering weeds.
And maybe even better : I think when there's a long-term topic like AI that requires more semiconductors and memory per unit of GDP. Charlie Munger said it best. He says: " Money is not made by buying and selling." "Money is made in waiting." So if there is a structural issue, it's best to just own the assets and forget about them.
Eventually, the AI story will become a bubble. This is a fact because whenever there is a structural demand story and volatility in the markets is underestimated, people make risky decisions because they underestimate the risk. But I don't think we're in the late stages of this AI bubble. The reason I don't think it's late stages is because once the Korean stock market crashed, most announced that it had peaked.
You know, if this was a bubble, people would say this is the bottom and start investing massively in the Korean market and semiconductors, but they didn't . They sold massively. So if I were to look at Cisco, okay, let's take a very short period from 1993 to 2000, okay? A 7-year period , but it was actually one cycle—the development of the Internet.
Cisco started at $80 and reached $9 in 1997 . So, this is a 10- fold movement, and then in 1997 there was a 40% correction. And in 1997, it happened during the Asian crisis. OK. So people would say: well, that's it , the peak has been reached, the Cisco story is over . Well, from $5 in 1998, it went up to $18.
Okay. That is twice as high as the previous maximum. And then in '98 we had, you know, Greenspan, irrational optimism, a few other events like the Russian default and the Long-Term Capital situation, and Cisco went from $18 to $9. OK. That is, it fell by 42%. And many have declared a peak in the technology sector.
In fact, I remember that period, there were many who danced on the graves of technology, claiming that this thing was over. Cisco rose from $9 to $ 80 by 2000 . So that was a 100- fold move from '93 to 2000, and that's a five-fold increase from its all-time high in 1998, which took just 18 months. That's when Cisco really peaked, because, as you correctly noted, the last orders were placed.
Keep in mind, in 2000, when I was a technical analyst and researcher, the peak was in 2000 because no one believed in the valuations at that point anymore. Cisco was worth 200 profits. At the heart of all this was the laying of fiber optics throughout the United States. These telecommunications companies (CLECs).
To justify the CLEC valuation, it would have been necessary to have a 10- year DCF with an exit multiplier of 30 and a cost of money of about 6%. Unrealistic assumptions, and they were the real buyers of Cisco equipment, right? They were real customers . It was a farce. So, if someone asks, is today's situation the same story?
This is not true. I mean, the number of people using AI is still quite small. He is already showing himself to be quite productive. The companies that buy this equipment, the hyperscalers, they are very, very serious players. These are not speculators trying to dig up land and sell rights to use infrastructure.
These are companies that actually buy, install racks, and receive large orders. Therefore, I believe that we are not yet in a "bubble" state. Tom Lee suggests that the crypto market is approaching a tipping point similar to the passage of the Securities Act of 1934, as the Senate advances the Clarity Act to a key vote to create uniform federal rules and clarify the jurisdictions of regulators.
While political hurdles remain, this legislative progress , complemented by joint guidance from the CFTC and SEC, could provide the regulatory clarity needed for large financial institutions to confidently scale tokenized assets and stablecoin infrastructure. At the same time, cryptoassets like Bitcoin and Ethereum have shown extraordinary resilience to traditional macroeconomic pressures such as inflation, high oil prices, and concerns about rate hikes.
Moreover, Ethereum is significantly ahead of Bitcoin, even despite fluctuations in legislative prospects and declining traders' expectations for a Fed rate hike amid weak labor market data. Ultimately, this ability of digital assets to strengthen despite macroeconomic and regulatory frictions serves as a “bullish” indicator, signaling strengthening crypto market fundamentals.
As Raoul Pal points out, the underlying global technological landscape is accelerating even faster. You know, I don't want to give you too long a history lesson, but Nvidia has proven that it has a sustainable revenue business in part because of the CUDA platform, and there's an almost guaranteed roadmap for people to buy upgrades.
Therefore, it should be revalued as a growth stock. I think the ratio should be closer to 25-30. Now about memory and semiconductor production equipment, which are two steps away from the end user - they have the risk of the so- called " bullwhip effect." This means that these industries can be more cyclical because they do not have full visibility into orders.
Let's say the end market is actually a consumer using AI, a lab service, like chat? GPT, OpenAI or DeepSeek, but they sign up through a lab that uses a hyperscaler, and that lab buys chips from Nvidia, but then Nvidia orders chips and other things from suppliers. Then the suppliers are far away from the customer, and there are a lot of double orders in the middle, right?
People are ordering because they think that memory and chip prices will increase; hyperscalers can double order in anticipation of price increases, they order now and then memory companies can create excess capacity in the future. This has actually always happened in every cycle. And, of course, there is a risk that it could happen this time too.
Therefore, cyclical groups typically have the lowest P/E ratios at the peak of the cycle. So, it is worth expecting the multiplier to decrease, but this is not a bearish signal. It just means you still want to bet on earnings preview. So I think there will still be a lot of earnings revisions . I mean, in the future, in a world of machine interaction, it's going to require a lot more memory and storage than a human, because a human consumes food and has a nervous system, and a robot needs memory and storage.
So I think the economy will become increasingly dependent on memory and semiconductors. So, the earnings review should be positive, but I wouldn't compare the memory multiplier to Nvidia. I think the answer to this question is ultimately not within our competence. That is, we already know today that AI is extremely capital-intensive to maintain, because the equipment wears out, and tokens are also spent to make it all work .
And open source models are cheaper, but partly because they are like generics, because they didn't have R&D costs, and many of them are distillation models. Ahem. These models are open source, but they cannot be completely free because someone has to pay for it . So one question we don't know the answer to is how these models are financed , and the second, as Elon Musk says, is that we are racing towards a singularity, where AI and robots can create so much productivity that everything will be free.
This is a kind of shock even for capitalism. So I think the answer is I don't know. This makes sense, I'm oversimplifying, but there's Linux and there's Windows, right? Yes. And there is Android and iOS. So the answer is—I don't know. But is this a negative for hyperscalers? You know, I don't think so .
These are again very serious companies, and they can choose a different path, as Apple did; Apple generates a lot of free cash flow, even though they had to endure criticism for not being AI-focused at the time. I think this is an existential question for investors that is very difficult to answer, so they should focus on where the opportunities exist.
Rahul Pal argues that the current economic and social instability reflects the rapid development of a new economic system driven by the exponential development of artificial intelligence and computing power, as evidenced by the investments of over $1 trillion in data center infrastructure by large technology companies.
In this era of exponential growth, traditional institutions are unable to keep up with the digital acceleration, creating the perfect foundation for blockchain networks and stablecoins designed for the continuous programmatic transfer of value. Pal predicts that by around 2030, an economic singularity will emerge , where autonomous AI agents will become major economic entities, requiring their own financial channels.
This shift is already happening thanks to initiatives like Google’s 2025 Agent Payments Protocol, supported by Coinbase and the Ethereum Foundation. Combining Pal’s vision of a machine-driven economy with Tom Lee’s forecast of regulatory clarity, expanding institutional investment, and the rise of stablecoins like USDC’s $73.3 billion in circulation, suggests that cryptoassets are evolving from speculative instruments to fundamental infrastructure for a transformed global financial system, even amid short-term market volatility. Friend.
Comments 0