WTF is Happening with South Korean Economy

Kind: captions Language: en Wall Street is panicking and the problem started 6,000 mi away. In the summer of 2026, more than 320,000 [music] South Korean investors saw their accounts liquidated in a matter of hours. Why? Because the world's most profitable company, [music] just did something it hasn't done in 20 years. Alphabet, Google's parent company, [music] burned $5.9 billion in a single quarter, all on AI. Big tech's AI obsession isn't an unlimited cash machine. It is a valued destroying black hole. When soul stops trading, New York is next and you will be in the fallout. For the last 5 years, the US NASDAQ 100 and the South Korean stock market, the Cosby, barely paid attention to each other. The 60-day rolling correlation average, how much they moved over time, stood at 0.16. It was pretty much just background noise. By mid 2026, it exploded, climbing to between 0.46 and 0.50. In other words, [music] the two markets are moving together, and it all comes down to one thing, AI. South Korea builds the hardware that keeps the US tech bubble [music] intact. Every night, Wall Street sleeps, and companies like Samsung and SKH Heinix open for trading. Their earnings, their guidance, their panic, all of it lands on trading screens of American investors when they finish their coffee. When US traders see major dips in Asia's hardware supplies, it dictates their own behavior. If those companies are in trouble, it sets off a chain reaction that impacts companies like Alphabet, who are trying and failing to reap the benefits of AI infrastructure projects. Korea's stock exchange is a monitor for the health of the entire AI sector. And that monitor has just flatlined. Behind it all is Google's 5.9 billion cash burn. It exposes a circular illusion, a closed loop where tech giants manufacture their own demand. Nvidia makes the physical chips that power nearly every major AI system on the planet. In 2025, Nvidia invested $100 billion into OpenAI. OpenAI used a large chunk of that money to buy Nvidia's chips. The cash goes out and the cash comes right back in. Nvidia's revenue looks strong. OpenAI looks well capitalized. Valuations go up, but the closed loop means that they don't even need paying customers to make the balance sheet look good. That's the important part for investors and shareholders. They see cash inflow and outflow and they see confidence in AI systems making them more likely to invest their own money into similar companies. That same problem isn't solely in investment but in the timing of returns. Unlike traditional investments where people would expect to see returns in a year or two, many AI investors believe the real payoff is three or even 5 years away. That hype leads to a second part of why AI investment is so rampant. FOMO. Investors want to get their foot in the door early. If they don't, someone else will end up reaping the rewards and that can be millions. Nvidia isn't alone in running this playbook. Microsoft poured billions into OpenAI and that money flowed straight back to Microsoft in the form of cloud computing contracts. It's the same loop under different branding. But Google's cash burn is what really happens when a company tries to compete in that environment without relying on circular deals. They try to absorb the cost of either building infrastructure or getting the necessary server load. Alphabet managed to spend $44.9 billion [music] in just 3 months. The entire premise was that it was banking on its own sustained growth to cover it. It was a record quarter for Alphabet. Revenue hit $119.8 billion, up 24%. Operating income climbed 30%. So, Google took those profits and they chose to burn billions in cash. The alternative is losing the AI race to Microsoft, Meta, and Amazon, who are all on their own AI spending spree. [music] It's not just one company making a gamble on AI. Every major player in the AI race is using some kind of cover up to spend more, but report [music] that it's spending less. If it pays off, it's hailed as a perfect strategy, a calculated gamble that came in. But while all that cash was disappearing into data centers, a second story was unfolding, and it made things in Korea much worse. A report surfaced that a homegrown Chinese company based in Shanghai was building its own deep ultraviolet lithography machines, the very equipment used to manufacture advanced semiconductors. According to the report, the company is planning to build five of these machines in 2026 and 20 more the following year. Most of the components are sourced domestically [music] with only a handful of parts still coming from Japan. Two weeks later, the company was named Shanghai Aishangna Electronic Technology [music] Group, also known as Aishangna, and it turned out to be majority state-owned. The news sent shock waves through the markets. The existing manufacturing was almost monopolized by ASML, based in the [music] Netherlands. It's the dominant supplier of both deep ultraviolet or DUV and extreme ultraviolet or EUV lithography machines. These are tools that actually etch the circuits onto the chips at a scale measured in nanometers. Out of the two, DUV is considered less technologically impressive, [music] but it still delivers results. But ASML has been barred from selling its most advanced EUV systems to China. Even if Aishanga makes a less advanced version domestically, it means China is suddenly free from the imposed export ban that keeps its huge industrial potential in check. When the report of Aishanga's existence came [music] out, Samsung Electronics and SKH share prices went down by more than 10% [music] in a single session. Tokyo Electron and TSMC dropped alongside [music] them, and the fear wasn't abstract. For years, the assumption was that IP restrictions and export controls kept China years behind on the equipment needed to make advanced [music] chips. Analysts estimated the gap at 7 to 10 years, but the market immediately reacted to the possibility that China might close that gap much faster than expected. But there's another side to this, one that has executives worrying. For years, Chinese firms have offered lucrative signing bonuses to entice engineers away from Samsung and other Korean chipmakers. They have been importing decades of institutional knowledge one hire at a time. Lee Bjongchu, a Siong Institute research fellow, has stated that Samsung and SKH DRAM technology has already been leaked to China. He also believes that a separate attempt to leak [music] HBM packaging technology was caught before it got out. That's only half of the problem. A civil war is erupting in South Korea's tech industry. Samsung is watching its chip engineers, the brains behind the hardware, walk out the front door to join SKH Highix. The insane AI profits have been weaponized with eyewatering bonus packages of up to $500,000 being offered to poach employees. The company that is supposed to be defending Korea's semiconductor edge against China is completely helpless as its own talent signs up with a domestic rival. It's not illegal and it isn't new. Corporate espionage, a talent war, and a Chinese domestic lithography program are now threatening South Korea's AI industry. And the global AI market relies on a country with no backup plan. If a cheap Chinese alternative starts to chip away at the monopoly, the entire financial foundation built by American AI giants collapses. But China isn't stopping there. It has another option, and it can flood the global market with cheap chips to crush South Korea. or it can ban chip exports entirely, fund their own AI industry, and steal a market of over a billion [music] people. For an industry that's running on pure speculation, that is a massive problem because AI isn't a finished product. Back in 2023, Alphabet's chairman, John Hennessy, told Reuters that a single exchange with a large language model costs [music] 10 times more than a traditional search. When the numbers were run, the added cost of AI searches could run to $6 billion a year. Google says those per query costs have fallen sharply since as much as 33 times less as hardware and models get more efficient. But this just leads to another problem. Reduced costs means the product gets used more often. [music] That means burning through more resources. Just an example of Jevon's paradox. [music] As technology becomes more efficient, we don't save resources, we just use them more. In 2025, Google's own environmental report admitted its emissions [music] had surged by 50% compared to 2019 before the AI trend had started. Tech giants are burning the world to protect [music] a product that has no clear purpose. According to an S&P global market intelligence survey, 42% of companies abandoned most of their AI initiatives in 2025. That's [music] more than double the 17% recorded the year before. On average, organizations scrapped nearly half of their AI proof ofconcept projects [music] before the projects ever reached production. Budgets got approved, pilots get built, demos got applause in the boardroom, and then nothing shipped. This pattern has a precedent, IBM. IBM spent years and billions of dollars building Watson Health into what was supposed to be a revolution in AI powered medicine. It was supposed to read scans faster than radiologists and cross reference treatment options. Instead, the project was sold off in pieces in 2022 before mainstream AI LLM grabbed the public's attention. Watson Health failed because despite having solid results in testing, no company could see how it would make them money. IBM also mandated that any prospective buyer abandon its existing tech [music] stack and adopt Watson's, essentially pigeon holing themselves for one solution. The trillion dollar valuations are all resting on a lie. The uncontrollable growth is based on investments leading to more investments. It's a corporate [music] carousel. So, who actually eats the loss when the leverage begins to unravel? Remember those Korean retail investors and those 360,000 liquidated accounts? Well, 62% [music] of them belonged to people under 35. These were young people who borrowed money because the market had been rising in a way that felt less like investing and more like a game everyone was winning. When the leverage snapped back, it snapped back on the people with the least capacity to absorb it. Some of them are now left owing brokers more money than the positions were ever worth. This affects America, too. A large share of ordinary retirement savings, the kind [music] sitting in index funds and 401ks, is concentrated in the same AI heavy companies burning cash at record rates. When you contribute to a retirement account without picking individual stocks, a portion of that money is flowing into the AI ambitions of a small group of companies, whether you like it or not. Passive investing was supposed to be the safe, boring [music] choice after the dot crash. The 2008 housing crash taught a generation not to gamble on individual names. Instead, the index itself became the concentrated bet. And there are pension funds. Public sector retirements, the accounts backing teachers, municipal workers, and union pensioners have been increasing their exposure to the same AI concentrated indices. These are some of the most consequential and least visible fault lines sitting inside the entire trade. The upside from AI [music] spending, if it ever fully materializes, gets captured by the companies making the bets [music] and the executives holding the stock options. If the bet goes wrong instead, the losses land on the people who never had a say in the decision and can least afford to take the hit. That's the [music] part that gets ignored every single time the story gets told in terms of stock tickers and index points. A crash isn't just numbers on a screen. It's a transfer of risk. One passed from the people who chose to take it to the people who didn't. The worst part, [music] this is not the first time. In August 2024, Japan's Nikay and Topics indexes [music] plunged more than 12% in a single session, the steepest 1-day drop since 1987. The trigger was a surprise rate hike from the Bank [music] of Japan that sent the yen surging. It forced a violent unwind of the so-called carry trade, where investors had been borrowing cheap yen to fund bets on assets like American tech stocks. When the yen strengthened, those investors had to sell to cover their positions, which pushed the yen higher still, triggering more forced selling. On the same day, Korea's cost dropped more than 8% and briefly halted trading. Go back further 25 years to the.com bubble, and the parallel gets sharper. [music] At the peak of that bottom, the top 20 companies in the S&P 500 made up 38% of the index's total value. In the 5 years before the crash, the Nasdaq had climbed 572%, pulling in investors who never had bought a stock before. When the crash came, the Nasdaq fell 78% over the space of 2 years, wiping out more than $5 trillion in market value. It didn't reclaim its old high for 15 years. Today, the top 20 companies in the S&P 500 account for more than half of the index's total value, with roughly 10 of them tied directly to AI and semiconductors. The reasoning investors give for holding on sounds almost identical to 2000. There are no meaningful earnings yet, but there's untapped potential in AI, and the companies investing in them are big on FOMO. There is one big difference this time. During the dotcom crash, the infrastructure that got built, the fiber optic cable laid by companies that went bankrupt, didn't just disappear. Cheap abandoned fiber ended up carrying the internet traffic of the next two decades. Someone paid pennies on the dollar for that fiber, and they built real businesses on top of it. But AI infrastructure doesn't leave that kind of gift behind. The chips inside these data centers physically degrade within 1 to 3 years, even losing most of their economic value the moment the next generation of hardware ships. After 3 years, a data center is full of outdated [music] chips and has ruined the local environment by soaking up water, emitting noise and pollution, and devastating the local real estate market. If the AI bet doesn't pay off, that data center is as good as scrap. The AI era doesn't leave a legacy, it ends with sky-high debt and a pile of useless tech that no one will want in 5 years. Put those two stories side by side, and it gets uncomfortable. In 2024, we saw how quickly a currency move on the other side of the world can force a same-day double-digit sell-off in soul. The year 2000 showed what happens when an entire market convinces itself [music] that the potential is the same thing as profit right up until the moment it isn't. 2026's crash carries pieces of the leverage and the speed of 2024 while having the concentration in the mispriced [music] potential of 2000. And this time it's unfolding when the two markets most linked to it, Korea's chipmakers and America's AI giants are more tightly bound together than at any point in the last 5 years. The question that actually matters isn't whether this ends. Bubbles always end. The question is who is still standing when the leverage finally runs out of places to hide and who gets handed the bill that they never agreed to pay. And China might be ready to pick up the pieces. To learn more, watch what China understands about AI that the US doesn't to see how it's ready to shake up the tech sector. Or click on this video.