Nvidia, SK Hynix Sign Multi-Year AI Memory Pact
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Nvidia and SK Hynix signed a multi-year co-development and supply agreement for next-generation memory chips, with SK Hynix set to provide memory for Nvidia's Vera Rubin platform and continue as Nvidia's largest memory partner across AI servers, PCs, and robotics.
- SK Hynix ordered $28.7 million in new equipment for next-gen HBM4 production, underscoring the concrete supply ramp tied to the deal.
- South Korea's science minister said the country will seek priority supply of Nvidia's Vera Rubin GPUs, while Nvidia also struck separate deals with Naver and Doosan to expand the country's broader AI infrastructure.
- Jensen Huang said the AI infrastructure boom will last over a decade and framed the broader AI stock sell-off as a buying opportunity — yet SK Hynix and Samsung shares closed with heavy losses on the day of the announcement.
- Multiple outlets framed the deal as a direct response to the chip industry's 'toughest bottleneck' (per The Next Web), with Tom's Hardware noting the agreement seeks to address extended development cycles for high-bandwidth memory.
Why it matters: The multi-year co-development pact locks SK Hynix in as the primary memory supplier for Nvidia's Vera Rubin platform, deepening integration between chip designer and memory maker just as high-bandwidth memory remains the AI industry's tightest bottleneck. Yet the market read it harshly: SK Hynix and Samsung shares closed with heavy losses, suggesting investors see near-term supply strain or margin pressure even as Jensen Huang framed the AI stock sell-off as a buying opportunity and projected a decade-long AI infrastructure boom.