Tom Lee Interview: This Round of KOSPI Plunge Is a Deleveraging Event, Avoid Trading Against the Structural Trend

Original Title: Tom Lee: "We are Close to the Bottom"
Original Source: Global Money Talk
Original Translation: DeepTech TechFlow

Disclosure: Tom Lee, Chairman of BitMine Immersion Technologies (BMNR), holds approximately 5.78 million ETH, making it the largest Ethereum holding institution globally. Lee also serves as the CIO of Fundstrat Capital, managing the $5 billion GRNY ETF, which holds positions in entities discussed in this article such as Robinhood. Lee's personal wealth is highly correlated to the price of ETH and the performance of GRNY. All views on Ethereum and the crypto market in this issue align with his significant financial interests. Furthermore, Fundstrat's core business model revolves around paid research subscriptions, and Lee's public statements serve a customer acquisition/marketing function. Readers are advised to consider these conflicts of interest.

Tom Lee is one of Wall Street's most ardent bulls. Fundstrat sells research monthly to hedge funds and family offices in 26 countries globally. The GRNY ETF, managed by him, has consistently outperformed the S&P 500 since launch. He is also the Chairman of BitMine, the world's largest ETH holding company. This recording takes place live at the NYSE amid the Korean Kospi's sharp monthly drop, a collective flash crash in AI semiconductor stocks, and debates in the market on whether the "AI bubble has burst." Lee's core argument is so simple it's almost blunt: the current sharp decline is due to leveraged funds being liquidated, not a fundamental turning point. He references Cisco's history from 1993-2000, where it underwent four deep corrections and eventually rose 100 times, believing that the current situation doesn't even qualify as late-stage AI bubble.

Key Points Summary

Tom Lee views the recent sharp drops in the Korean stock market and AI semiconductor sector over the past month as a "forced deleveraging" event. The Korean market has introduced a large number of leveraged products in recent years, amplifying two-way volatility. According to data from a U.S. bulge bracket broker, hedge funds' selling of tech stock long positions is at nearly a decade's record pace; the "weak hands" have been washed out. He cites two famous quotes from Peter Lynch and Charlie Munger to support his core advice: don't trade in structural trends, "money is made by sitting, not trading."

Lee uses Cisco's history from 1993-2000 to illustrate AI's current position: Cisco experienced multiple pullbacks of over 40% during that period, with each announcement that "tech stocks were finished," but eventually rose from $0.80 to $80, a 100-fold increase. The key difference was that the peak in 2000 was a true bubble: buyers were unrealistic fiber optic companies rationalizing purchases with impractical DCF models, whereas "today's buyers are hyperscalers, serious about buying equipment, shelving, and taking large orders." Regarding the impact of Chinese AI models (Kimi K3), Lee acknowledges it as an existential issue "beyond my pay grade," but notes that open-source models are fundamentally like generics; someone's R&D costs will always need to be covered. He is more focused on downstream AI opportunities (Mag 7, software, crypto) starting to outperform upstream semiconductors, and his GRNY ETF is thriving on sticking to this framework, outperforming 92% of peers this year. For the macro outlook in the second half of the year, Lee bets that inflation will be lower than expected (peak oil price impact has passed, housing and wages are weakening), which will compel the Fed to turn dovish.

Highlights Summary

Deleveraging Pressure, Not the End of the Story

“The speed at which hedge funds have been selling off tech stock long positions is the fastest in nearly 10 years. The weak hands have already been shaken out.”

“Every time the market surges straight up, leveraged longs get caught, and then they get liquidated. That's what's happening now.”

“No one can perfectly time the market bottom. But if you sell now, exit, and wait for a signal to re-enter, you'll end up chasing at a higher level.”

Trimming the Flowers: Don't Day Trade in a Bull Market

“Peter Lynch once said, selling your winners is like cutting the flowers and watering the weeds.”

“Charlie Munger said it best: money is not made through buying and selling, money is made by sitting and waiting.”

“If there's a structural theme, you should buy into it and then forget it.”

NVIDIA's P/E at 16x is Not Expensive, But Memory Stocks Are Naturally More Cyclical

“NVIDIA's forward P/E is 16x, not over 20x. It has a CUDA moat and an almost certain upgrade roadmap, which should be revalued as a 25-30x growth stock.”

“Memory and semiconductor equipment are two layers away from end customers, with a risk of bullwhip effect: hyperscalers might double-order due to expected price increases, and memory manufacturers could over-expand in the future.”

“Cyclical stocks have the lowest P/Es at the peak of the cycle, which does not mean it's a sell signal. You just need to bet that earnings estimates will continue to rise.”

Cisco Halved Four Times on Its Way to 100x: AI Is Not in the Late Stage Yet

“Cisco rose 100-fold from 1993 to 2000. It was halved at least four times in between, with people each time saying tech stocks were done for.”

“If this is indeed the late stage of the AI bubble, people should be shouting ‘this is the bottom, quickly buy semiconductor stocks.’ But what are they doing? They are crazily selling out.”

“In 2000, Cisco had a P/E of 200x, and the buyers were fiber optic companies doing a 10-year DCF at a 6% discount rate. Today's buyers are hyperscalers, not hippies digging up and selling fiber optics.”

Ethereum vs Bitcoin: Yield Asset vs Digital Gold

「Bitcoin is a value store, with the ecosystem aiming to 'freeze' it as digital gold. Ethereum is a yield asset, with a staking yield of around 3%.」

「BitMine's current staking yield is approximately $6 million per week, $3 billion per year. If ETH reaches $5,000, this number would be close to $1 billion annually.」

「Our perpetual preferred shares only require a $30 million dividend payment per year, which can be easily covered solely by staking rewards.」

Crypto Catalysts in Line: CLARITY Act + Robinhood Chain + Institutional Onboarding

「Since the end of June, Ethereum has outperformed the semiconductor sector by 72 percentage points. Some have lost 40% on semiconductors but gained nearly 30% on Ethereum.」

「Robinhood Chain is built on Ethereum, not on any other chain. The daily trading volume has already exceeded $1 billion, and Robinhood could potentially earn $1 billion in just one year from this chain alone.」

「The CLARITY Act has now been put in the queue; it will establish a single regulatory body for the entire crypto economy. Similar bills have already been passed in Japan and Russia, and the U.S. must catch up.」

Gold Is Still Valuable, Just Taking a Breather from Its Rally

「Gold's price surge over the past 3 years is at a 5 standard deviation level in the metal's 12-century history. It certainly needs to digest that.」

「In an AI-driven world, gold's hedging function as a store of value won't change. I recommend everyone to hold some, maybe 1%.」

South Korea Plunge: Leverage Products Amplify Volatility, but the Story Remains

Host: On June 22nd, the South Korean Kospi touched nearly 9,300 points, while the Philadelphia Semiconductor Index also peaked on the same day. The past month has been a true plunge. People are wondering: Why the sudden impulse rally followed by a sharp reversal? Are we following Kospi's lead, or is there global concern about AI trading overheating?

Tom Lee: South Korea has performed exceptionally well in recent years, not just in 2026, but in several years prior. The underlying logic is that the amount of semiconductors and memory used per unit of global GDP output is continuously rising. This means that South Korea, as an economy and stock market, will be much more important in the next ten years compared to the past 30 or even 50 years. Profits should perform very well."

However, over the past few quarters, the Korean market has introduced a large number of leveraged products, which has amplified the two-way volatility. When the market sharply rises, leveraged long positions get trapped, and then they get liquidated. This is what is happening now: a forced deleveraging event. But this does not mean that the underlying story is over. So I believe this pullback will prove to be one of the best buying opportunities for semiconductor and AI stocks. Consequently, the Korean stock market, AI stocks, memory stocks, and semiconductors will eventually reach new highs significantly above the previous ones.

Host: It's hard to believe until you see it, but what signal are you waiting for to confirm that the pullback is over? Or is now the opportunity, and should one enter in tranches?

Tom Lee: First of all, timing the market has never been profitable. If you hold these stocks, you should continue to hold. If you sold and exited, and are now waiting for a signal to enter, you will ultimately end up chasing at a higher level. No one can perfectly time the bottom.

However, the signal you want to see (massive deleveraging) has already occurred. If you look at the data from U.S. prime brokers, the speed at which hedge funds have been unwinding their tech stock long positions is the fastest in almost three years, possibly the fastest in nearly a decade. They have already gone through a round of massive deleveraging. We have also seen some very high-profile forced liquidation stories in Korea. So, if anyone has been forced to sell ("weak hands"), they are already out. I assess that we are quite close to the bottom.

But people will still make mistakes: trying to time the top, trying to time the bottom. In reality, the people who make the most money are those who hold onto their positions. Peter Lynch once said a very famous line: "Selling your winners and holding your losers is like cutting the flowers and watering the weeds." Charlie Munger put it even better: "Money is not made in buying and selling; money is made in waiting." If this is a more structural theme involving more semiconductors and memory, you should buy and forget about it.

NVIDIA 16x PE vs. Memory Stocks 4.5x PE: Not a Simple Comparison

Host: I remember you said in a previous conversation, "Shorts sound smart, but longs make money." I want to ask a valuation question. What is NVIDIA's forward PE ratio? I think it's around twenty-something times?

Tom Lee: It's actually 16 times.

Host: 16 times. Well, that's indeed... Their earnings expectations are very strong. On the other hand, SK Hynix was around 7 times at its peak and has now dropped to 4.5 times. Can you directly compare these two?

Tom Lee: NVIDIA has proven itself to have a certain degree of recurring revenue, as it has the CUDA platform and almost a definite upgrade roadmap that keeps people buying. It should be revalued as a growth stock, and I think a reasonable valuation would be between 25 and 30 times.

What about memory and semiconductor equipment? They are two layers away from the end customer, facing the risk of the bullwhip effect. In simple terms, these industries are more cyclical because they lack pure order visibility. Let's assume the end market is a consumer using some AI lab service, such as ChatGPT or DeepSeek. He subscribes through an AI lab that uses a hyperscaler, the hyperscaler buys chips from NVIDIA, and NVIDIA then places orders with suppliers. The suppliers are too far from the end user. In this process of layers of transfer in between, there will be a lot of duplicate orders. If the hyperscaler expects memory and chip prices to rise, they may order double the amount in advance to lock in prices. Then the memory manufacturers may overproduce in the future.

This has happened in every cycle, and of course, there is a risk it will happen this time. So, the more cyclical the sector, the lower the PE at the cycle top, which is normal, and you should expect the PE to compress. But this is not a sell signal; you just need to bet on earnings forecasts continuing to rise. In a machine-to-machine world, robots need much more memory and storage than humans: humans eat, have a nervous system, robots need memory and storage. The economy is becoming more and more memory-intensive and semiconductor-intensive. So I believe earnings forecasts will continue to rise. But don't compare the PE of memory with that of NVIDIA.

Cisco's History Lesson: Slashed Four Times, Eventually Up 100x

Host: You mentioned earlier that you didn't want to give a long history lesson, but I think your memory of many historical events is very useful for investors. I remember when Cisco ran into trouble, it was to some extent a bullwhip effect: in 2001, orders suddenly collapsed because people had previously placed too many repeated orders, falling like a row of dominoes. Does this lesson apply now, or is it too early?

Tom Lee: The AI story will eventually turn into a bubble; this is inevitable. Whenever there is a structural demand story, and the market underestimates volatility, people make risk-adjustment mistakes: underestimating risk.

But I don't believe we are in the late stages of an AI bubble. The reason is simple: the stock market has just started to fall, and most people are already calling the top. If it were really a bubble, people should be saying "this is the bottom" and then crazily investing in semiconductors. But they aren't; they are selling like crazy.

Let's take a look at Cisco. From 1993 to 2000, a span of 7 years, but really just one cycle: the Internet buildout cycle. Cisco started at $0.80. By 1997, it had surged to $9, a 10x increase. Then in 1997, it retraced by 40% amidst the Asian financial crisis, and everyone was saying, "Cisco's story is over." But what happened? By 1998, it had climbed from $5 to $18, doubling its previous high. Then in 1998, another setback with Greenspan's "irrational exuberance" speech, the Russian default, and the Long-Term Capital Management collapse. Cisco fell from $18 to $9, a 42% drop. Many then declared the end of the tech stock era. I vividly remember that time, with many dancing on the grave of tech stocks, proclaiming the end of this trade. Then Cisco climbed from $9 all the way to $80 by 2000. From 1993 to 2000, it increased by a total of 100x. And from the previous high in 1998, it took just 18 months to rise by 5x.

That was truly Cisco's peak. In 2000, I was a tech analyst. Why was that peak real? Because nobody believed the valuation anymore: Cisco had a 200x PE ratio. The fundamental issue was that the companies laying fiber (CLECs) were Cisco's true buyers, and to justify the CLEC valuation, you needed a 6% cost of capital, a 30x exit multiple, and a 10-year DCF. These assumptions were completely unrealistic. It was a farce. If someone asks if today is the same story, it's not. Today, the number of people using AI is still relatively low, but AI has already shown significant productivity. The companies (hyperscalers) buying these devices today are very serious companies. They are not hippies here to dig dirt and sell IRUs; they are actually purchasing equipment, putting it up, and securing large orders. So I believe we are far from a bubble stage.

Kimi K3 Impact and the China Model: A Matter Beyond the "Pay Grade"

Host: When it comes to AI, we've just experienced the "Kimi moment," likened to the DeepSeek moment. This Chinese open-source model has 28 trillion parameters, not as cost-effective as some models, but still impressive. If the Chinese model can reach close to the level of American models, will it slow down investments in companies like OpenAI and Anthropic? Or will the demand shift towards Chinese models?

Tom Lee: To be honest, the answer to this question is beyond my pay grade. What we already know is that AI is extremely capital-intensive; just the maintenance is very costly, equipment will age, and there's also token consumption. Open-source models are indeed cheaper, but part of the reason is they are like generics: no R&D costs, many are distilled models. These models are open-source, but they can't truly be free; someone always foots the bill.

The second level, as Elon Musk put it, we are heading towards the "Singularity." AI and robots could create such immense productivity that everything becomes almost free. This is disruptive even to capitalism itself. So the answer is: I don't know. Just like there is Linux and Windows, Android and iOS, this logic holds. But is this negative for hyperscalers? I don't think so. These are all very serious companies. They can choose not to participate, just like Apple did in the past. Apple thus generated a lot of free cash flow, but also faced criticism for not being AI-forward enough. For investors, this existential question is difficult to answer. They should focus more on where the opportunities lie.

Outlook for the Second Half of the Year: Betting on Lower-Than-Expected Inflation, AI Outperformance in the Downstream

Host: Speaking of opportunities, the first half of the year has already ended. What is your outlook for the second half? July has historically been favorable for the stock market, but this year's performance has been mixed due to events surpassing salary levels.

Tom Lee: Our strategy has been effective throughout this year. Our Granny Shots ETF (GRNY) has outperformed the S&P 500 by about 120 basis points year-to-date, ranking in the top decile among its peers, outperforming over 92% of fund managers. The reason we have been able to achieve this is because we stick to long-term themes: AI downstream, cybersecurity, and Fed monetary easing. Even as the market turns hawkish, we are still betting on the Fed shifting to a dovish stance. Small-caps have performed exceptionally well this year, which is actually the Fed's biggest signal of tilting dovishness.

We haven't made too many adjustments for the second half of the year. Earnings growth is accelerating, we are in the midst of earnings season, and the AI narrative is very compelling. However, we are willing to buy downstream: the Mag 7, software, and crypto. These are part of the AI downstream narrative, and they have already started to outperform.

Regarding the Fed, the bond market is currently hawkish, believing that the Fed must raise rates. Our bet is that inflation will be lower than expected. People are overly focused on oil as an inflation driver, and the oil price impact has already occurred; I believe the influence of oil prices on inflation has peaked. The underlying drivers of inflation are weakening: housing is soft, and wages are not truly accelerating. This will ultimately position the Fed to be able to cut rates.

Host: I completely agree. I don't think this Fed really has the intention to raise rates. Your GRNY fund has now reached $5 billion in size, hasn't it?

Tom Lee: To be precise, it's close to $5 billion.

Host: Great, I hold some myself. Maybe I should just go all into GRNY instead of trying to time the market.

Tom Lee: When AI and memory stocks were soaring, people criticized our fund for not having a heavy allocation to AI and semiconductors. We had exposure, but not a heavy allocation. As a result, when memory and AI saw a 40% pullback, our fund outperformed due to anchoring in the longer-term idea of AI trades.

BitMine vs MicroStrategy: Why ETH Treasury Companies Are a Different Ball Game

Host: I'd like to switch gears and talk about BitMine. It's an Ethereum treasury company, possibly the first and most well-known of its kind. MicroStrategy was the first Bitcoin treasury company, but they went through some very public struggles. What sets BitMine's approach apart?

Tom Lee: First, a clarification: BitMine is the world's largest Ethereum holding institution, with approximately 5.78 million ETH, making it the largest ETH holder globally. However, we are not actually the first Ethereum treasury company; we are probably the fourth or fifth, but we have become the largest. We have just crossed the one-year mark in operation.

We have three core differences from MicroStrategy.

First, the underlying asset. Bitcoin is a store of value, with the Bitcoin ecosystem aiming to "harden" it, avoiding changes and always remaining digital gold. Ethereum, on the other hand, is different. It is a yielding asset, with a staking yield of about 3%. Ethereum's ecosystem is constantly evolving and is the largest ecosystem in the crypto space, even bigger than Bitcoin's. The entire ecosystem's goal is to make Ethereum the Wall Street financial settlement layer, with the entire financial track ultimately operating on top of Ethereum-based stablecoins.

Second, while MicroStrategy took a relatively passive approach to Bitcoin—holding and creating digital credits—BitMine deeply engages with the Ethereum ecosystem. We led investment rounds in three companies that split off from the Ethereum Foundation, all focused on strengthening Ethereum (improving price or enhancing the ecosystem). We are actively shaping the future of Ethereum.

Third, Balance Sheet Complexity. MicroStrategy intentionally designed its balance sheet to be complex: it has convertible bonds, four classes of preferred stock, and common stock, with these components sometimes competing with each other. Since Bitcoin does not have native yield, they have to sell stock to pay dividends. BitMine's capital structure is extremely simple: it only has common stock and recently issued perpetual preferred stock. The current staking revenue is approximately $6 million per week and around $3 billion per year. If ETH rises to $5,000, the annual staking revenue will be close to $10 billion. The perpetual preferred stock only needs to pay $30 million in dividends annually. The staking revenue alone can easily cover this.

So you should view BitMine as a company deeply embedded in the Ethereum ecosystem. We are betting on Ethereum to become not only Wall Street's settlement layer but also the settlement layer for communication between robots.

Host: So, is there a significant difference between buying ETH tokens itself and buying BitMine stock?

Tom Lee: If you can directly buy ETH and stake it yourself, that's fine. But there are two types of investors who cannot do that. The first type is institutional investors: asset management companies that manage a huge amount of funds cannot hold ETH tokens directly because it requires maintaining a crypto wallet. However, they can buy stocks. BitMine has been included in the Russell 1000 large-cap stock index and trades on the NYSE. It is currently the only large-cap Ethereum stock available for major fund managers to buy. The Russell 1000 is the largest and most widely used benchmark index. If you manage a large fund in Boston and want Ethereum exposure, you can only buy BitMine.

The second type includes investors who want to use options and derivatives or those looking for higher ETH exposure. BitMine outperforms ETH during uptrends and has a thriving options and perpetual contract market. The stock investor world is $240 trillion, while the native crypto investor world is only a few hundred billion. Betting on the stock world coming to buy BitMine may be a better choice.

The Catalysts for Crypto Are Lining Up, But the Current Quietness May Signal the Bottom

Host: The crypto market has experienced a significant pullback and is now almost in a state of being forgotten, with few participants and little discussion. However, looking at Bitcoin's chart, it seems poised to break a quite extended downtrend. Will this quietness end soon?

Tom Lee: What you are describing is the look of a bear market. In a bear market, nobody talks about stocks. When Apple was falling, nobody talked about Apple either. Price drives emotion, and in a pullback, everyone is naturally bearish at the bottom. This quietness is exactly what the bottom formation mechanism looks like: deleveraging and resetting expectations.

But there are plenty of crypto catalysts. Since the end of June, Ethereum has outperformed the stock market by 72 percentage points. Some people lost 40% in stocks but gained nearly 30% in Ethereum. The CLARITY Act has just "moved to the one-yard line," which will establish a single federal regulatory agency for the entire crypto economy. Currently, the U.S. is regulated by each state, leading to fragmented rules. Japan has already passed a similar version, and Russia has just done so too. The U.S. must catch up. This will usher in the era of institutional adoption of cryptocurrency, a market larger than anything in crypto history.

If you've been in the crypto space for a long time, what you've experienced is the "hobbyist phase," what I call Ethereum 1.0, the era of meme coins and NFTs. The future market is stablecoins and payment rails. Robinhood wants to tokenize everything. They could choose any blockchain to build on, but they chose Ethereum and launched Robinhood Chain. This has already been a breakthrough success: the daily trading volume exceeds $1 billion, and Robinhood could potentially earn $1 billion just from this chain in a year, which is huge for them. Every company on Wall Street is watching what Robinhood has done and realizing that tokenizing assets on Ethereum can make a lot of money.

Host: I noticed that Robinhood is one of the key holdings in your GRNY fund.

Tom Lee: Yes.

Host: Speaking of bear markets that no one is discussing, precious metals are in the same boat. Six months ago, everyone was talking about gold and silver, and now no one is asking. What's your take?

Tom Lee: Gold's 3-year return is roughly 5 standard deviations above its 12-century average. It certainly needs to take a breather. Maybe there's another 10% downside, but that's about it. At Fundstrat, we've always recommended allocating a portion to gold, like 1%. Because whether it's debt uncertainty, AI's impact on society, or social unrest, gold's safe-haven function as a store of value won't change. This is especially important in an AI-driven world. So, I think everyone should hold some gold, but it has gone up too much and needs to take a break.

Original Article Link