Semiconductor Supercycle

Origin

The term 'supercycle' originated in commodity markets. As China's industrialization drove a decade-long rise in oil, iron ore, and copper prices during the 2000s, analysts labeled it a 'supercycle' transcending ordinary business cycles. Goldman Sachs's Jeffrey Currie popularized the concept in the early 2020s by forecasting a new commodities supercycle based on supply underinvestment.

The semiconductor industry borrowed this term to describe structural long-term booms beyond the typical 3–4 year boom-bust cycle. The fundamental mechanism is: expanded capex during the boom → oversupply 18–24 months later → price collapse. Supercycle proponents argue this pattern won't apply this time. But history has repeatedly proven otherwise.

Major Supercycles

2017–2018 Memory Supercycle: Cloud data center migration and high-spec smartphones drove a DRAM price surge. South Korea's semiconductor exports exceeded $9 billion for three consecutive months, and Samsung's semiconductor division posted over 24 trillion won in operating profit through Q3 2017. By Q4 2019, DRAM prices had fallen 40% from their peak.

2020–2021 Pandemic Supercycle: Remote work and online education triggered PC, tablet, and data center investment, but inventory buildup and cooling demand pushed the industry into a downcycle by 2022.

2024–2026 AI Supercycle: The emergence of generative AI and exploding HBM demand are driving the current supercycle. Optimists argue that AI's cumulative data demand and supplier-customer co-design represent a 'structural shift.' Yet the simultaneous capex expansion by Samsung, SK hynix, and Micron may already be planting the seeds of oversupply for 2027–2028.