SK Hynix Chairman: Wafer Capacity to Triple by 2034, Memory Shortage May Persist Until 2030
Alina Collins
SK Hynix Chairman Chey Tae-won says wafer capacity will double within five years and triple by around 2034 — yet warns that memory-chip shortages driven by AI demand could persist until 2030.
What does tripling capacity actually look like?
SK Hynix is building four chip factories in South Korea simultaneously; the first phase of one is set to come online next year.
Chey's timeline: capacity doubles within five years and triples by around 2034.
This means → this is not a normal expansion cycle — it is a full rebuild of the production base. A construction plan originally scheduled to run through 2045 has been pulled forward.
Triple the output and still not enough?
Chey's own words: "Once all these facilities are built, capacity won't just double — it will triple. But people are already saying even that won't be enough."
Earlier this month in Taipei he added: the sustained shortage could last until 2030.
In plain terms = AI data centers are consuming memory chips faster than factories can be built. The supply-demand gap has no visible end date.
Is SK Hynix alone in this view?
The global memory market is dominated by SK Hynix, Samsung Electronics, and Micron Technology. Nvidia ranks among SK Hynix's largest customers.
Micron's CEO recently judged that supply tightness will extend beyond 2026 — echoing Chey's assessment.
This reflects a collective bet by the top memory makers that AI demand is a long cycle, not a short-term spike.
Where does the U.S. listing stand?
SK Hynix confidentially filed with the U.S. SEC earlier this year.
The company told investors that market feedback on the plan has been "extremely positive."
This means → massive expansion requires massive capital. A U.S. listing opens a funding channel and deepens the supply-chain tie to American customers — Nvidia above all.
Content is for reference only, not financial advice.