AI Boom Anxiety Triggers 10% Kospi Crash on Chip Stock Dump

For AI professionals, this crash is a direct referendum on the industry's hardware bill. The sell-off in HBM memory leaders Samsung and SK Hynix signals that investors are questioning whether the astronomical capex in AI infrastructure will translate into sustainable revenue. If the markets are right, the era of endless chip orders may be ending—and with it, the free-flowing funding that has fueled AI R&D.

South Korea's benchmark Kospi index suffered a dramatic intraday collapse on July 28, 2026, plunging more than 10% in one of its steepest single-day declines since the global financial crisis. The rout was concentrated in the country's massive chipmaking sector, with heavy selling of Samsung Electronics and SK Hynix shares triggering circuit breakers and sending shockwaves through Asian equity markets. The trigger: growing unease over whether the artificial intelligence boom, which has powered semiconductor demand to unprecedented heights, can sustain its torrid pace.

“

By the close, the Kospi had shed roughly $150 billion in value, with Samsung alone losing over $30 billion in market cap.

The sell-off reflects a broader reckoning over AI's return on investment. For two years, the promise of generative AI drove hyperscalers to pour hundreds of billions of dollars into data center infrastructure, lifting South Korean memory giants—dominant suppliers of high-bandwidth memory (HBM) essential for AI accelerators—to record valuations. Samsung and SK Hynix accounts for roughly 30% of Kospi's market capitalization, meaning any rotation out of chip names has an outsized effect on the index. By the close, the Kospi had shed roughly $150 billion in value, with Samsung alone losing over $30 billion in market cap.

Broader Asian markets also felt the tremors. Japan's Nikkei 225 fell 3%, Hong Kong's Hang Seng dropped 2.5%, and Taiwan's TAIEX, another chip-heavy index, lost 4%. However, the Kospi's 10%+ plunge stood out as a clear rejection of the AI premium that had been built into chip stocks. The sell-off was exacerbated by algorithmic trading and margin calls, as leveraged positions in tech shares were forcibly unwound.

The proximate cause: a series of analyst notes and industry reports questioning whether AI model builders are seeing sufficient revenue growth to justify ongoing capex. In recent weeks, Chinese AI upstart DeepSeek demonstrated that high-performance models can be trained with fewer chips, challenging assumptions of insatiable demand. Additionally, global memory chip prices, which had surged on AI demand, showed signs of peaking. Samsung's preliminary Q2 earnings, released earlier in July, hinted at slowing HBM order momentum, even as overall numbers remained strong.

For investors, the crash underscores the fragility of a market narrative built on AI's limitless growth. The Kospi's price-to-earnings ratio had ballooned to over 20x, well above historical averages, fueled by expectations of ever-rising AI chip orders. Now, with fear overtaking greed, the market is repricing risk. South Korean policymakers and financial regulators convened emergency meetings, but there is little they can do to counter a global sentiment shift.