Why Tech Stocks Are Falling: KOSPI Crash, Fed Rate Fears and AI Trade Reset Explained

South Korea's KOSPI closed down 10% today, its fifth-worst session in stock market history. The Nasdaq Composite is down another 2.5% pre-market. Nvidia, Intel, AMD, Micron, Alphabet, Broadcom, Teradyne, and Western Digital are all down between 5% and 9% before US markets open.  

The obvious narrative circulating is that this is the Broadcom story still playing out which has caused AI and tech stocks to bleed. But it isn't. That explanation is three weeks stale and points at the wrong mechanism entirely.

Let's break down what is actually driving the AI & tech stocks selloff today, why the KOSPI is falling harder than any other major index, and what the distinction between a valuation reset and an AI trade collapse means for how investors should think about this.

Why KOSPI Crashed Today: The 3 Triggers of Doom

Samsung stock and SK Hynix stocks fell up to 12% today, dragging KOSPI's 10% lower. This index collapse has direct, Korea-specific causes instead of generic "AI selloff" narrative.

1. MSCI did not add South Korea to its Developed Markets watchlist in its annual review. This was the key bull catalyst for Korean equities in 2026, as investors were betting on future passive inflows from developed-market funds. NH Investment & Securities had called watchlist inclusion a high-probability outcome on June 18. With no upgrade now possible before June 2027, the thesis weakens, especially after foreign outflows of over $78 billion this year.

2. SK Hynix’s production shift is being misread. Reports said the company may curb HBM output and allocate more capacity to conventional DRAM. The market saw this as weaker AI demand. But SK Hynix’s 2026 HBM supply was already sold out, and Q1 operating margin stood at a record 72%. Conventional DRAM ASPs rose mid-60% in Q1 due to shortages, so shifting capacity toward higher near-term margins looks like rational allocation, not an AI demand slowdown.

3. The KOSPI rally had become stretched. The index was up roughly 80-90% year-to-date, briefly making South Korea the world’s sixth-largest equity market. Samsung and SK Hynix also crossed $1 trillion in combined value. Hana Securities warned that SK Hynix overtaking Samsung’s market cap could mark a bull-market peak, comparing it to Cisco’s dot-com-era peak in 2000. That bubble comparison is now adding pressure.

Samsung and SK Hynix together represent approximately 42% of the KOSPI by market cap. When these three forces hit simultaneously,  the resulting selloff is not a verdict on whether AI chip demand is structurally intact. It's a forced unwinding of a highly concentrated, highly appreciated position that lost two of its main supporting narratives on the same morning.

How Fed Rate Fears Are Pressuring AI, Tech Stocks

The broader Nasdaq selloff is a result of US Fed interest rate decision expectations. What is actually compressing AI stock prices is not a single earnings report from a single chipmaker. It is a shift in the rate environment that has been building since the June 17 Fed meeting, and it operates on AI valuations regardless of whether the underlying demand for chips and cloud is strong.

A high-growth stock like Nvidia does not fall only when its business weakens. It can also fall when interest rates move up.

Think of it this way: Investors are paying today for profits Nvidia may earn many years from now. When a stock trades at around 32 times forward earnings, the market is valuing not just next year’s profit, but several years of future growth.

Now, when rates rise, those future profits become slightly less valuable in today’s money. Because when interest rates rise, investors can earn more from safer options like government bonds. So they become less willing to pay a premium today for profits a company may generate many years later.

That is why a 0.50% rise in rates by the Fed can reduce the present value of five-year future earnings by roughly 12-15%. This does not mean Nvidia’s AI business is weak. The company can still be growing revenue at 78% year-on-year, guiding for $91 billion in quarterly revenue, and dominating AI chips, yet the stock valuation can still fall.

That is what is happening. The June 17 Fed meeting revealed that nine of eighteen officials now see interest rates rising in 2026. New Fed Chair Kevin Warsh has been consistently hawkish.

Bank of America strategist Michael Hartnett had warned that June could be difficult for equities, and that risk is now playing out. His key point is simple: inflation is still too high, while the job market is not weak enough for the Fed to cut rates. Since 1960, this setup has happened only six times: 1966, 1973, 1990, 2000, 2008 and 2021. Each period involved Fed rate hikes and was difficult for stocks.

Hartnett also noted that when CPI stays above 4%, the S&P 500 has historically fallen around 4% over the next three months and 7% over six months. He compares today’s setup with 1994, when the Fed had to hike rates aggressively after strong jobs data. Stocks then moved sideways until bond yields finally stopped rising.

The next key trigger is today’s PCE inflation print. Economists expect it to rise to 4.1%, from 3.8% in April. A hotter number would support Bank of America’s three-rate-hike view and keep pressure on AI stocks, because higher rates make future earnings less valuable today.

Did SpaceX’s Stock Crash Trigger the AI Selloff?

Another trigger is the SpaceX stock price crashing over 25% in three days. SpaceX’s post-IPO fall shows that investors are becoming less patient with expensive, loss-making AI stories, and are no longer willing to fund unlimited losses at premium valuations.

This also matters for existing AI stocks. SpaceX has already listed, while Anthropic and OpenAI have confidentially filed for IPOs. Together, these mega-listings could absorb over $200 billion of public market capital in 2026.

For years, investors bought Nvidia, Microsoft, and Google as indirect AI plays. But once Anthropic and OpenAI list, investors can buy the AI model companies directly. That could reduce demand for existing AI proxies and act as a structural headwind for stocks that previously benefited from limited AI investment options.

The Market Is Selling the AI Proxy, Not the Product

The most important piece of data in today's sessions is not the KOSPI level or the Nasdaq futures. It is the fact that Micron Technology rose 6.82% on Monday June 22, the same day Alphabet fell 5%, Amazon fell 4.8%, Microsoft fell 3%, and SpaceX collapsed 16%. 

That 12-point divergence between two stocks both described as "AI plays" on the same day tells you exactly what kind of market this is. It is not one trade. It never was one trade, and the investors treating it as one are the ones getting hurt.

Not all AI stocks are being hit for the same reason.

Alphabet’s issue is company-specific. Losing top AI talent like Noam Shazeer and John Jumper, to rival labs is a real concern. It suggests that Alphabet may be facing pressure in the race for frontier AI talent. D.A. Davidson analyst Gil Luria summed it up clearly: the frontier AI race now appears to be between Anthropic and OpenAI.

Micron’s strength points to the opposite trend. Its HBM4 shipments for Nvidia’s Vera Rubin platform are on track. For fiscal Q3 2026, the market expects around $34.5 billion in revenue and 81% gross margins. If CEO Sanjay Mehrotra confirms that Micron’s full 2026 HBM output remains locked under long-term contracts, it would directly challenge the idea that AI infrastructure demand is slowing.

Bottom Line: AI Trade is Not Collapsing, Just Resetting?

The AI trade is not unravelling yet. In fact, it looks like a selective reset, and not a dot-com-style collapse. In 2000, the market sold almost everything. Today, some AI names are falling while others are holding up or rising. Here’s how the market is being reset in two ways.

• Higher rates are making future profits less valuable today. That hurts high-growth AI stocks, even if their businesses remain strong.
• AI investing is becoming more fragmented. Vanda Research calls this shift the move from the Magnificent 7 to the FAB 10: Frontier AI and Big Tech. This adds SpaceX, Anthropic and OpenAI to the existing seven. These new names could absorb capital that earlier had to flow indirectly into stocks like Nvidia, Microsoft, Alphabet and Meta.

Two events, Micron Earnings on Wednesday and PCE on Thursday, will tell you whether the pressure is demand-side (AI orders slowing) or purely rate-side (strong earnings, compressed multiples). 

If it's the latter, the long-term AI infrastructure thesis remains intact, and the current selloff reflects a re-rating event rather than a fundamental breakdown. Those are two very different things, and they warrant very different responses.