SK Hynix and Nvidia expand partnership to develop next-generation AI memory
SK Hynix secured a multiyear co-engineering agreement with Nvidia to develop memory systems for AI infrastructure, the companies announced June 8. The deal extends an existing partnership and covers memory for Nvidia's Vera CPUs, Spark desktop supercomputers, and Jetson Thor robotic computing platforms.
The partnership addresses a critical bottleneck in AI infrastructure buildout: memory supply cannot keep pace with demand. SK Hynix, one of three major memory manufacturers, has pushed lead times as far back as 2028 and delayed its high-bandwidth memory (HBM4) production to the third quarter of 2026 from the second quarter.
Both companies will focus on reducing development cycles and tackling advanced fabrication challenges. SK Hynix Chairman Chey Tae-won said the pair have "been building toward this for years," framing the agreement as formalization of ongoing work rather than a new direction.
Using AI to accelerate chip design and manufacturing
SK Hynix is already applying Nvidia software to speed its own operations. The company uses Nvidia's CUDA-X libraries and AI to accelerate semiconductor simulations for computational lithography workflows, and its PhysicsNeMo framework to speed AI physics calculations.
SK Hynix also uses Nvidia's Omniverse platform to create digital twins of its manufacturing facilities, allowing engineers to test optimizations before implementing them on the factory floor.
Market context
SK Hynix's market capitalization surpassed $1 trillion for the first time this year, driven by unprecedented demand for AI memory. The company holds certification to supply Nvidia with HBM4 products alongside two other major manufacturers.
Nvidia CEO Jensen Huang said at Computex that all three certified suppliers "are in production, and they're all racing to support Vera Rubin." Earlier reports suggested HBM4 delays would impact Vera Rubin's rollout, but Nvidia has denied those claims.
For IT professionals and developers building AI systems, understanding the hardware supply chain and memory architecture matters. Learn more about AI for IT & Development or explore AI learning paths for software engineers to understand how infrastructure decisions affect application design.
Your membership also unlocks:
