Chip factories are massive industrial operations that run continuously and use huge amounts of electricity. When energy prices jump, the cost of running those factories rises quickly.
A stock market crash in South Korea is raising a startling question about the future of artificial intelligence. What if the biggest risk to the AI revolution is not chips or software but the fossil fuel energy that powers the factories making them?
South Korea’s KOSPI index plunged about 15 percent in just 48 hours, wiping nearly 270 billion dollars off the market. Circuit breakers halted trading for the first time in 576 days. Semiconductor giants Samsung Electronics and SK Hynix led the selloff, falling roughly 10 percent and 12 percent.
At first glance the reason looked simple. Rising tensions in the Middle East pushed oil prices above 80 dollars a barrel and sparked fears about shipping through the Strait of Hormuz, a narrow sea route through which a huge share of global oil travels.
But investment analyst Shanaka Anslem Perera says the crash may be revealing something deeper about the AI boom.
“Everyone sees a Korea crash,” Perera wrote in a note on X. “Almost nobody sees the AI supply chain crisis hiding inside it.”
To understand why, start with how AI actually works
Modern AI systems depend on powerful chips such as the NVIDIA Blackwell GPU, the NVIDIA H100 GPU, and custom processors like the Google Tensor Processing Unit. These chips train AI models and run massive data centers.
But those processors cannot function alone. They need extremely fast memory to feed them data.
That memory is largely produced in one country.
Samsung and SK Hynix together control about 67 percent of the world’s DRAM memory and almost 80 percent of high bandwidth memory revenue. High bandwidth memory, often called HBM, is essential for AI chips.
Perera described it in simple terms.
“HBM is the oxygen of every AI data center being built on Earth right now,” he wrote.
That means the AI boom depends heavily on factories in South Korea.
Those factories depend heavily on energy.
South Korea imports about 97 percent of the energy it uses. A large portion of its oil and liquefied natural gas comes from the Middle East and travels through the Strait of Hormuz.
That narrow waterway is only about 21 miles wide at its tightest point.
If conflict threatens ships moving through the strait, energy prices can surge or supplies can slow. For a country that imports almost all of its fuel, that creates immediate pressure.
Perera says that is why the recent market crash matters.
“The KOSPI crash is not a Korea story,” he wrote. “It is the first live stress test of the AI infrastructure buildout’s most critical single point of failure.”
Chip factories are massive industrial operations that run continuously and use huge amounts of electricity. When energy prices jump, the cost of running those factories rises quickly.
If the shock lasts long enough, companies may delay expansion plans or reduce production.
That matters because the world has very little memory stockpiled.
Perera estimates global DRAM inventories cover only two to three weeks of supply, while NAND flash inventories cover roughly three to four weeks.
“There is no buffer,” he warned. “If Hormuz disruption persists beyond a month, production cuts become unavoidable.”
Financial markets may already be reacting
While semiconductor stocks dropped sharply, defense companies surged. Shares of Hanwha Aerospace jumped about 20 percent and LIG Nex1 climbed roughly 30 percent.
Perera says that shift reveals what investors are starting to realize.
“Capital is not fleeing Korea,” he wrote. “It is rotating from the idea that energy is a solved problem into the idea that energy is the binding constraint on everything.”
The effects could spread far beyond South Korea
Countries across Asia, including India, depend on Middle Eastern oil and gas. Higher energy prices can push up inflation, weaken currencies and raise the cost of running data centers.
That means building AI infrastructure could become more expensive or slower than expected.
Perera says the short term outcome still depends on geopolitics.
“If conflict resolves within ten days and oil returns below seventy five dollars, this was the buying opportunity of the year,” he wrote.
But the crash revealed a structural risk investors had largely ignored.
“The AI supercycle has a chokepoint,” Perera warned. “It is not chips. It is not talent. It is not capital.”
“It is the energy that powers the fabs that make the memory that makes AI possible.”
And much of that energy flows through one narrow strait.
