Nvidia certifies all three memory makers for Vera Rubin HBM4, SK Hynix takes the lion's share
Jensen Huang confirms Samsung, SK Hynix and Micron passed HBM4 qual on June 5; SK Hynix holds ~60-70% of Rubin volume, 12-layer stacks priced above $600
Summary
On 5 June 2026, Nvidia CEO Jensen Huang confirmed that Samsung, SK Hynix and Micron all passed certification to supply HBM4 for the next-generation Vera Rubin accelerator, which entered full production after the 1 June GTC Taipei keynote. All three qualifying matters less than the split: analysts put SK Hynix at roughly 60-70% of Rubin HBM4 volume, Samsung at ~25-30%, and Micron supplying the remainder. HBM4 runs near $500/stack, with 12-layer parts expected above $600, a step up from HBM3E. First customer shipments are scheduled for this summer, broad availability in H2 2026. The allocation cements Korea's grip on the highest-value node of the AI stack.
By the numbers
- 3, memory makers certified for Rubin HBM4 (Samsung, SK Hynix, Micron); qual confirmed June 5.
- ~60-70%, SK Hynix share of Rubin HBM4 volume.
- ~25-30%, Samsung's share; Micron supplies the balance.
- ~$500, HBM4 price per stack; 12-layer parts expected above $600.
- June 1, Vera Rubin full production announced at GTC Taipei.
Why it matters
HBM4 is the gating component of every Rubin GPU, and the supplier split decides who captures the richest slice of the accelerator's bill of materials. SK Hynix's structural lead, rooted in yield, not just qualification, keeps the AI-memory rent concentrated in Korea, and underpins its market-cap overtaking of Samsung. Micron's slice is the US foothold in an otherwise Korea-led market.
What to watch
- Whether Samsung or Micron claws share off SK Hynix as Rubin volume scales.
- HBM4 pricing trajectory and 12-layer mix through H2 2026.
- HBM4E samples and the move toward in-house 2nm base dies (Samsung's stated lever).