Nvidia CEO Says SK Hynix's Plan to Double Wafer Capacity by 2030 Is Not Enough

aiai-infrastructureai-compute-chips 2 posts · 2 accounts

Nvidia's chief executive said SK Hynix's plan to double wafer capacity by 2030 is not enough. The comment comes after SK Hynix outlined plans to lift DRAM wafer input capacity to about 1 million wafers a month by 2030 to 2031 from roughly 550,000 now.

The South Korea expansion is centered on the Yongin semiconductor cluster and the Cheongju M15X fab, with the new capacity slated to be DRAM. Air Liquide has committed $233 million to support the buildout, which comes as demand tied to AI chips rises.

From the sources (2 posts)

@cointelegraph

🇰🇷 AI: AI chip demand is surging as Air Liquide commits $233M to support SK Hynix’s expansion in South Korea.

@firstadopter

I said this last week

Market impact

42 medium

MU NVDA SOXX

Why it matters: Nvidia's CEO saying even a doubling of SK Hynix DRAM/HBM wafer capacity by 2030 is insufficient signals structurally tight memory supply against AI demand, giving pricing-power read-through to memory makers (MU proxy for SK Hynix/Samsung) and reinforcing the AI-compute thesis for NVDA and semis.

At publish time: Headline quotes the Nvidia CEO calling a planned capacity doubling "not enough" — a demand-outstrips-supply signal for HBM/DRAM from a pivotal buyer, tradable on the facts alone.

model confidence: medium

What happened next

1–2 weeks later

played out moved up

MU rallied ~11% to record highs within two weeks on sold-out HBM and rising DRAM/NAND prices; tight-memory pricing-power thesis for MU/NVDA/semis played out.

based on follow-up coverage

Follow-up coverage

No follow-up coverage matched this story.

Intel Foundry Seen Profitable in 2H27 as 18A Yields Near 80% Russia’s Seaborne Diesel and Gasoil Exports Drop 18% in May as Drone Attacks Curb Loadings