Dialogue with Tom Lee: The Recent Plunge in Korean Stocks is a Forced Deleveraging Event; Do Not Trade the Swings in Structural Trends

Compiled & Edited by: Deep Tide TechFlow

Guest: Tom Lee, Co-founder and Head of Research at Fundstrat Global Advisors, CIO of Fundstrat Capital (managing the GRNY ETF, ~$5B AUM), Chairman of the Board of BitMine Immersion Technologies (BMNR)

Host: (Global Money Talk, recorded live at the NYSE)

Podcast Source: Global Money Talk

Original Title: Tom Lee: "We are Close to the Bottom"

Air Date: July 27, 2026

Conflict of Interest Disclosure: Tom Lee is Chairman of BitMine Immersion Technologies (BMNR), which holds approximately 5.78 million ETH, making it the largest institutional holder of Ethereum globally. Lee also serves as CIO of Fundstrat Capital, managing the GRNY ETF (~$5B), which holds positions in companies discussed in this episode, such as Robinhood. Lee's personal wealth is highly correlated with the price of ETH and the performance of GRNY. All views expressed regarding Ethereum and the crypto market in this episode align with his significant financial interests. Additionally, Fundstrat's core business model is paid research subscriptions, meaning Lee's public commentary serves an acquisition/marketing function. Listeners are advised to consider these relationships when evaluating his statements.

Tom Lee is one of Wall Street's most steadfast bulls. His firm, Fundstrat, sells research monthly to hedge funds and family offices across 26 countries. The GRNY ETF he manages has consistently outperformed the S&P 500 since its launch, and he is Chairman of BitMine, the world's largest corporate ETH holder. This episode was recorded live at the NYSE, against the backdrop of a monthly crash in South Korea's Kospi index, a flash crash in AI semiconductor stocks, and a market debating whether the "AI bubble has burst." Lee's core argument is almost brutally simple: the current sell-off is a forced liquidation of leveraged positions, not a fundamental shift in the narrative. He points to Cisco's history of dropping 50% four times between 1993 and 2000, only to eventually rally 100x, arguing that we are not even in the late stages of the AI bubble.

Key Takeaways

Tom Lee believes the crash in the Korean stock market and AI semiconductor stocks over the past month was a "forced deleveraging" event. The Korean market has seen a significant introduction of leveraged products in recent years, amplifying two-way volatility. Data from US investment bank prime brokers shows that the speed at which hedge funds have been selling long positions in tech stocks has set a record in nearly a decade; the "weak hands" have been flushed out. He cites famous quotes from Peter Lynch and Charlie Munger to support his core advice: do not trade in and out of secular trends. "Money is made by sitting, not by trading."

Lee uses the history of Cisco from 1993-2000 to illustrate AI's current position. Cisco experienced multiple 40%+ drawdowns during that cycle, with each one prompting declarations that "tech is dead," but it ultimately went from $0.80 to $80 – a 100x gain. The key difference, he argues, is that the top in 2000 was a real bubble: buyers were fiber-optic companies using unrealistic DCF models to justify purchases. "Today's buyers are hyperscalers, serious companies buying equipment, racking it, and landing big orders." Regarding the impact of Chinese AI models (like Kimi K3), Lee admits it's an existential question that is "above his pay grade," but points out that open-source models are essentially generics; someone still has to pay for the R&D. He is more focused on the observation that AI downstream opportunities (Mag 7, Software, Crypto) are starting to outperform upstream semiconductors, and his GRNY ETF has beaten 92% of its peers this year by sticking to this framework. For the macro outlook in the second half of the year, Lee is betting that inflation will come in lower than expected (the oil price shock has peaked, housing and wages are weakening), which will force the Fed to pivot dovish.

Highlights of Key Insights

Forced Deleveraging, Not the End of the Story

"Hedge funds are selling long tech positions at the fastest pace in nearly a decade. Weak hands have been shaken out."

"Every time the market goes straight up, it traps leveraged longs, and then they get margin-called. That's what's happening right now."

"No one can perfectly time the bottom. But if you sell now, wait for a signal, and then buy back, you'll end up chasing it at higher levels."

Sell the Flowers, Water the Weeds: Don't Trade in a Bull Market

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

"Charlie Munger said it better: Money is not made by buying and selling; money is made by sitting and waiting."

"If there's a structural theme, you should buy and forget about it."

NVIDIA at 16x P/E is Not Expensive, But Memory Stocks are Inherently More Cyclical

"NVIDIA's forward P/E is 16x, not in the twenties. It has the CUDA moat and a near-certain upgrade roadmap. It should be re-rated as a growth stock at 25-30x."

"Memory and semiconductor equipment are two layers removed from the end customer. They face bullwhip effect risk: hyperscalers might double-order due to expected price increases, and memory manufacturers could over-expand capacity later."

"Cyclical stocks have their lowest P/E at the top of the cycle. That's not a sell signal. You just need to bet that earnings estimates will continue to go up."

Cisco Dropped 50% Four Times and Then Rallied 100x: AI is Not in the Late Stages

"Cisco rallied 100x from 1993 to 2000. It had at least four 50% drawdowns in between. Each time, people said tech was finished."

"If this were really the late stages of the AI bubble, people would be shouting 'This is the bottom, buy the dip in semiconductors!' But what are they doing? They're panic selling."

"In 2000, Cisco traded at 200x P/E. Buyers were fiber companies using a 6% discount rate on a 10-year DCF. Today's buyers are hyperscalers. They are not hippies digging ditches to lay fiber."

Ethereum vs Bitcoin: Yield-Bearing Asset vs Digital Gold

"Bitcoin is a store of value; the ecosystem wants to 'ossify' it into digital gold. Ethereum is a yield-bearing asset with a staking yield of about 3%."

"BitMine's current staking yield is roughly $6 million per week, or $300 million annually. If ETH goes to $5,000, that figure approaches $1 billion annually."

"Our perpetual preferred stock only requires $30 million in dividends per year. We can cover that with staking yield alone."

A Queue of Crypto Catalysts: CLARITY Act + Robinhood Chain + Institutional Entry

"Since the end of June, Ethereum has outperformed memory stocks by 72 percentage points. Someone down 40% on memory could be up nearly 30% on Ethereum."

"Robinhood Chain is built on Ethereum, not another chain. Daily volume has already surpassed $1 billion. Robinhood could potentially make $1 billion a year from this chain alone."

"The CLARITY Act is on the one-yard line. It would establish a single federal regulator for the entire crypto economy. Japan and Russia have already passed similar legislation. The US must catch up."

Gold Isn't Less Attractive, It Just Rallied Too Much and Needs a Break

"Gold's rally over the past 3 years is a 5-standard-deviation event over its 12-century history. It naturally needs to digest those gains."

"In an AI world, gold's function as a store of value and safe haven doesn't change. I recommend everyone hold some, perhaps 1%."

Korea's Crash: Leveraged Products Amplified Volatility, But the Story Remains

Host: On June 22nd, the Korean Kospi hit nearly 9,300 points, coinciding with the peak of the Philadelphia Semiconductor Index. The past month has been a true crash. People are asking: why the spike up first, then this sudden reversal? Are we following the Kospi, or is there a global fear that the AI trade is overheated?

Korea has performed remarkably well over the past few years, not just 2026. The underlying logic is that the amount of semiconductors and memory used per unit of global GDP is constantly increasing. This suggests that Korea, as an economy and stock market, will be far more important over the next decade than it has been over the last 30 or even 50 years. Earnings should be good.

However, in the past few quarters, the Korean market introduced a significant amount of leveraged products, which amplified two-way volatility. When the market rallies straight up, it traps leveraged longs, and then they get margin-called. That's what's happening now: a forced deleveraging. But this doesn't mean 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. By extension, the Korean stock market, AI stocks, memory stocks, and semiconductors will eventually make new highs much higher than before.

Host: It's hard to believe before you see it, but what signal are you waiting for to confirm the pullback is over? Or is this the opportunity, and one should just start dollar-cost averaging in?

First, timing the market is never beneficial. If you hold these stocks, you should continue to hold. If you sold and are now waiting for a signal to get back in, you will likely end up chasing the market at higher levels. No one can perfectly time the bottom.

That said, the signal you want to see (massive deleveraging) has already happened. If you look at the prime brokerage data from US investment banks, hedge funds have been selling long tech positions at the fastest pace in three years, maybe even the fastest in a decade. They have already gone through a massive deleveraging. We've also seen some high-profile forced liquidation stories in Korea. So, if someone was a forced seller ("weak hands"), they are out. My judgment is that we are fairly close to the bottom.

But people still make mistakes: trying to pick tops and bottoms. In reality, the people who make the most money are the ones who stay invested. Peter Lynch famously said: Selling your winners is like cutting the flowers and watering the weeds. Charlie Munger said it better: "Money is not made by buying and selling; money is made by sitting and waiting." If this is a secular theme involving more semiconductors and memory, you should buy and forget about it.

NVIDIA at 16x P/E vs Memory Stocks at 4.5x: A Direct Comparison Is Not Simple

Host: I remember you saying in a previous conversation: "Bears sound smart, but bulls make money." I want to ask about valuation. What is NVIDIA's forward P/E? I think it's around the mid-twenties?

It's actually 16x.

Host: 16x, okay... their earnings expectations are very strong. Conversely, SK Hynix, which was around 7x at its peak, has now dropped to 4.5x. Can you directly compare the two?

NVIDIA has proven it has a degree of recurring revenue, thanks to the CUDA platform and a near-certain upgrade roadmap that compels people to keep buying. It should be re-rated as a growth stock; I think a reasonable valuation is between 25 and 30 times.

Memory and semiconductor equipment are two layers removed from the end customer, exposing them to the risk of the bullwhip effect. Simply put, these industries are more cyclical because they lack pure order visibility. Suppose the end market is a consumer using an AI lab service like ChatGPT or DeepSeek. They subscribe to an AI lab, which uses a hyperscaler, which buys chips from NVIDIA, which then orders from suppliers. The supplier is too far from the end user. In the chain of transmission, a lot of double-ordering can occur. If hyperscalers expect memory and chips to get more expensive, they might place double orders now to lock in prices. Then, memory manufacturers might over-expand capacity later.

This has happened in every cycle, and there's a risk it happens again. So, the more cyclical the stock, the lower its P/E is at the cycle peak. That's normal; you should expect P/E compression. But it's not a sell signal. You just need to bet that earnings estimates will continue to be revised upwards. In a machine-to-machine world, robots will need far more memory and storage than humans: humans eat and have nervous systems; robots need memory and storage. The economy is becoming more memory and semiconductor intensive. So I think earnings estimates will continue to be revised up. But don't compare the P/E of memory stocks to NVIDIA's.

History Lesson from Cisco: Dropped 50% Four Times, Eventually Rallied 100x

Host: You said you didn't want to give a long history lesson, but I think your memory of certain historical events is very useful for investors. I remember when Cisco got into trouble, it was partly due to the bullwhip effect: orders collapsed in 2001 because people had been double-ordering so much beforehand, it fell like dominoes. Is that lesson applicable now, or is it too early?

The AI story will eventually become a bubble; that's inevitable. Anytime you have a story driven by structural demand, and the market underestimates the volatility, people will make poorly risk-adjusted decisions: they underestimate the risk.

But I don't think we are in the late stages of the AI bubble. The reason is simple: As soon as the stock market started to fall, the majority declared it was the top. If it were truly a bubble, people would be saying "This is the bottom" and frantically pouring money into semiconductors. But they aren't; they are frantically selling.

Look at Cisco. From 1993 to 2000, a 7-year period that was really just one cycle: the internet buildout cycle. Cisco started at $0.80. By 1997, it rallied to $9, a 10x gain. Then in 1997, it corrected 40% during the Asian Financial Crisis, and everyone said, "Cisco's story is over." What happened? By 1998, it went from $5 to $18, doubling its previous high. Then in 1998, more trouble: Greenspan's "irrational exuberance" speech, the Russian default, and the Long-Term Capital Management collapse. Cisco dropped from $18 to $9, a 42% decline. Again, many people declared tech stocks had topped. I vividly remember that period; many were dancing on the grave of tech stocks, saying the trade was finished. Then Cisco rallied from $9 all the way to $80 by 2000. From 1993