Semiconductor company shares rocketed 70% in the first half of 2026. The industry's best starting six months on record. Artificial intelligence was the driver: as companies built data centers and trained AI systems, demand for specialized memory chips surged. Micron Technology shares climbed over 870% year-on-year, according to Reuters.

Company earnings have backed up the price surge. Samsung reported operating profit of 57.2 trillion won in the first quarter of 2026—eight times higher than the year before. Taiwan's TSMC, which manufactures advanced chips for companies like Google and Amazon, posted a 58% profit increase in 2026 after major cloud providers rushed to place orders, per CNBC.

But there is something important to understand about how concentrated this success is. TSMC does not just make a lot of chips. It makes roughly three-quarters of the world's most advanced chips. This company alone represents 41.5% of Taiwan's entire stock market. Samsung and SK Hynix, a South Korean memory chip maker, together account for more than half of South Korea's main stock index. When three companies own that much weight in their countries' stock markets, what happens to the chip business directly determines what happens to entire economies' investment values.

The demand for all these chips comes from a handful of American cloud companies—Amazon, Google, Microsoft. If those companies decide to slow down their AI spending, the ripple effect would spread quickly. Chipmakers' profits would shrink. Stock prices would fall. And because chipmakers dominate the stock indexes in Taiwan and South Korea, those countries' entire equity markets would take a hit.

Here is what we know and don't know. Cloud companies have committed to record spending on AI infrastructure in 2026 and into 2027. Equipment makers like ASML see continued strong order visibility. But stock price gains of 70–80% in six months leave little cushion for disappointment. The market has priced in years of sustained AI growth. If new data suggests that growth is slower than expected—if companies built data centers before they actually needed them, or if AI software becomes more efficient and requires fewer chips—then the repricing would be sudden and severe.