News

NVIDIA and SK Hynix Sign Multi-Year Memory Deal Spanning Rubin to Robotics

NVIDIA and SK Hynix announced a multi-year technology partnership on June 7 to co-develop next-gen memory for NVIDIA's AI factories, covering everything from the Vera Rubin supercomputer to the Jetson Thor robotics platform. NVDA rose ~2.3% in pre-market trading on June 8.

OurAlpha

09 Jun 2026 — 2 min read

NVIDIA locks in memory supply across its full product stack — from AI supercomputers to robotics to AI PCs.

TL;DR

On June 7, NVIDIA (NVDA) and South Korean memory chipmaker SK Hynix announced a multi-year technology partnership to co-develop next-generation memory for NVIDIA's "AI factories." NVDA rose ~2.3% in pre-market trading on June 8.

Platforms covered: NVIDIA's next-gen Vera Rubin AI supercomputer, Vera CPU, RTX Spark AI PC, and Jetson Thor robotics compute module^{[NVIDIA IR]}
Memory types: HBM4, LPDDR5X, and 3D NAND
Scope: Beyond supply, both companies will apply AI to chip design and fab operations — including digital twins and automated production lines
Stock: NVDA +~2.3% pre-market on June 8^[Schwab]

On June 7, NVIDIA (NVDA) and SK Hynix announced a multi-year technology partnership to co-develop next-generation memory for what NVIDIA calls its "AI factories." The deal spans multiple product lines and extends into using AI to optimize chip design and fab manufacturing. NVDA climbed ~2.3% in pre-market trading on June 8, lifted by the news and a broader semiconductor rebound.

What's Covered: Platforms and Memory

Per the official joint announcement, the two companies will co-develop memory for the following NVIDIA platforms^[^{SK hynix Newsroom}^]:

Vera Rubin AI Supercomputer: HBM4, LPDDR5X, 3D NAND
Vera CPU: LPDDR5X
RTX Spark AI PC: LPDDR5X, 3D NAND
Jetson Thor Robotics Compute Module: LPDDR5X, 3D NAND

According to Tom's Hardware, the agreement is structured as a "memory co-development and supply" deal, with one stated goal being to address the lengthening development cycles that have become a bottleneck in advanced memory production^[^{Tom's Hardware}^].

Beyond Supply: Putting AI to Work in the Fab

The partnership also brings AI into the semiconductor design and manufacturing process itself^[^{NVIDIA IR}^]:

NVIDIA's CUDA-X libraries and PhysicsNeMo will be used to accelerate semiconductor simulation
SK Hynix will deploy NVIDIA Omniverse, OpenUSD, and cuOpt to build factory digital twins, targeting higher levels of fab automation

The announcement frames the collaboration around the three markets NVIDIA is building toward: AI infrastructure, personal AI, and physical AI. SK Hynix will serve all three — supplying HBM4 for Vera Rubin, LPDDR5X for the RTX Spark AI PC, and LPDDR5X and 3D NAND for the Jetson Thor robotics module.

Stock Move and Sector Context

Per Schwab's pre-market recap, NVDA rose ~2.3% before the open on June 8. The broader semiconductor sector bounced back after Friday's selloff, with memory and AI chip names including Micron (MU) and Marvell (MRVL) also trading higher^[^Schwab^].

One note for US investors: SK Hynix is listed in South Korea and is not a US-traded equity. The directly investable names tied to this news are NVIDIA and related memory supply-chain companies. Also on the calendar this week: the US May CPI print drops on June 10.