Nvidia and SK Hynix Sign Multi-Year Deal on Next-Generation AI Memory

*Nvidia and SK Hynix will jointly design future memory chips aimed at artificial-intelligence workloads.*

Nvidia Corp. and SK Hynix Inc. have entered a multi-year agreement to develop the next generations of memory chips for AI systems. The pact gives the South Korean company a direct role in shaping the high-bandwidth memory that Nvidia’s accelerators require.

The deal formalizes an existing supply relationship. SK Hynix already leads production of HBM3 and HBM3E stacks used in current Nvidia GPUs. Under the new arrangement the two firms will co-design subsequent memory architectures rather than treating SK Hynix solely as a foundry partner.

No financial terms or specific product timelines were disclosed. The announcement positions SK Hynix as the primary memory supplier for Nvidia’s upcoming AI platforms, reinforcing its standing among the handful of companies able to produce advanced HBM at scale.

Why it matters

For Nvidia the agreement reduces the risk of memory shortages that have constrained GPU shipments. For SK Hynix it locks in long-term volume commitments and gives its engineers early visibility into Nvidia’s bandwidth and power requirements. Other memory makers now face a clearer technology roadmap set by the Nvidia–SK Hynix pairing rather than open industry standards.

---

Sources:

• Bloomberg Technology

{
  "excerpt": "Nvidia and SK Hynix agree to co-design future high-bandwidth memory chips for AI accelerators.",
  "suggestedSection": "business",
  "suggestedTags": ["nvidia", "sk-hynix", "hbm", "ai-hardware"],
  "imagePrompt": "Abstract view of stacked silicon wafers and copper interconnects on a dark workbench, soft reflections on metallic surfaces, faint circuit traces visible under low light. muted color palette, cinematic lighting, 16:9"
}