Nvidia, SK Hynix In Multiyear Deal To Develop AI Memory Chips; SK Hynix Stock Drops
Nvidia said it signed a multiyear partnership with SK Hynix to co-develop next-generation AI memory for “AI factories” and to apply AI in chip design and manufacturing. The deal targets memory supply for Nvidia’s Vera Rubin, Vera CPUs, RTX Spark PCs and Jetson Thor platforms, using CUDA-X and PhysicsNeMo. SK Hynix shares fell 7.68% to 1,911,000 won.
Deal reinforces Nvidia’s AI-factory roadmap and could support demand visibility for advanced memory used in its systems.
Nvidia signed a multiyear partnership with SK Hynix to co-develop next-gen AI memory and use Nvidia CUDA-X/PhysicsNeMo for chip workflows.
Near-term upside bias for NVDA on deal credibility; magnitude likely limited without disclosed financial terms.
Background
The article frames the deal as addressing extended development cycles for advanced AI memory and adds a software/digital-twin layer (CUDA-X, PhysicsNeMo, Omniverse/OpenUSD/cuOpt).
Why it matters
For Nvidia, the partnership can improve alignment of memory supply with its AI infrastructure roadmap. For SK Hynix, the strategic win is offset by a sharp same-day equity decline, implying the market is focused on timing, economics, or memory-cycle fundamentals.
Market relevance
A concrete multiyear co-development agreement links Nvidia’s AI systems to SK Hynix’s next-gen memory and manufacturing digital-twin tooling, with an immediate mixed market reaction.
Market effects
Strengthens the AI-factory supply chain narrative (compute + advanced memory) and highlights tighter co-design between system vendors and memory makers.
Korea memory complex may see sentiment spillover, though the reported SK Hynix selloff signals investors are not uniformly positive.
Supports global AI hardware capex expectations and could influence how traders price advanced memory development cycles.
Alternative perspectives
Without disclosed volumes/pricing, the partnership may be viewed as incremental R&D rather than near-term earnings support—explaining SK Hynix’s drop.
Investors may be discounting extended development cycles and execution risk for next-gen AI memory, despite the broad platform integration (Vera/RTX Spark/Jetson).
Key entities
- companyNvidia
Co-develops next-generation AI memory with SK Hynix and applies CUDA-X/PhysicsNeMo and Omniverse-related tooling in the workflow.
- companySK Hynix
Co-develops advanced AI memory for Nvidia platforms and builds factory digital twins using Nvidia Omniverse/OpenUSD/cuOpt.
- product/platformVera Rubin
Nvidia AI supercomputers referenced as a target platform for the co-developed memory.
