NVIDIA and SK hynix announced a multiyear technology partnership on June 7, 2026 (U.S. time) to jointly develop next-generation memory for AI factories. The two will collaborate across design and manufacturing to advance HBM4 and beyond and to expand supply.
June 7, 2026 · NVIDIA × SK hynix
A Multi-Year Pact to Feed the AI Factory With Memory
NVIDIA names SK hynix its "largest memory partner," locking in next-generation HBM and DRAM supply across the Vera Rubin, Vera CPU, RTX Spark and Jetson Thor lineups — and applying AI to chip design itself.
50–70%
SK hynix share of the HBM market — the leading supplier
2027
DRAM supply for Vera CPU confirmed through
~$13B
New fab investment to bolster SK hynix capacity
HBM Market — Goldman Sachs projection
SK hynix expected to retain a majority grip amid competition with Samsung and Micron.
Each block ≈ 10 percentage points of HBM share
Why memory is the bottleneck
HBM heavily shapes AI accelerator performance and tends to throttle the GPU. Stable supply is the lever for scaling AI data centers.
Secure HBM supply
→
Unblock GPU performance
→
Scale AI factories
What's locked in
Multi-year partnership, effective immediately
Targets: Vera Rubin, Vera CPU, RTX Spark, Jetson Thor
AI-driven design via CUDA-X & PhysicsNeMo digital twins
Still unannounced
Generation specs (HBM4 / GDDR7), pricing, benchmarks
Mass-production timing
Hands-on user evaluations — reactions remain market-level
▲ ~2%
NVIDIA shares after the news
▼ then ▲
SK hynix dipped, then recovered
With memory shortages possibly lasting years, the deal aims to stabilize the supply chain against sustained AI demand.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…