Nvidia and South Korea’s SK Group announced a partnership worth more than $500 billion on July 24, 2026, in San Francisco. The deal secures a long-term supply of high-bandwidth memory (HBM) from SK Hynix and includes the development of large-scale AI data centers expected to be operational by 2027.
The scale of the agreement is a massive bet on the physical constraints of artificial intelligence. By locking down $500 billion in memory supply, Nvidia is moving to eliminate the “bottleneck” that has historically limited how many AI accelerators the company can ship.
The HBM Supply Chain and Market Control
High-bandwidth memory (HBM) is the specialized DRAM that feeds data to GPUs.
The market for this technology is highly concentrated. SK Hynix controls approximately 58% of the HBM market, while Samsung and Micron split roughly 21% each. By securing a half-trillion-dollar commitment, Nvidia isn’t just ensuring its own growth; it is potentially starving competitors like AMD and Intel, who rely on the same limited supply chain for their respective accelerators.
This strategic alignment extends into future hardware. The partnership focuses on jointly developing high-bandwidth memory for AI training, AI agents, and physical AI applications, specifically targeting next-generation variants like HBM4.
SK Telecom’s 2-Gigawatt Infrastructure Buildout
The deal isn’t limited to chips. It includes a massive infrastructure push where SK Telecom will build a cloud business utilizing Nvidia’s Vera Rubin systems. The target capacity for this buildout is 2 gigawatts of power, a scale that implies the installation of hundreds of thousands of GPUs.
The first of these facilities is expected to come online in 2027. These data centers will be powered by a combination of Vera Rubin chips and SK Hynix’s HBM4 memory, ensuring that the hardware and the memory supporting it are developed in lockstep.
Korean Strategic Investments: Naver and Hyundai
Beyond the SK Group partnership, Nvidia is diversifying its footprint in South Korea. The company announced a $1 billion investment into Naver, a Korean cloud company. This project aims to expand Naver’s AI data centers, providing 200 megawatts of capacity for customers to secure AI computing power.
Nvidia is also branching into “physical AI.” According to The Korea Times, Jensen Huang announced plans to jointly develop a self-driving Genesis sedan with Hyundai Motor Group. This follows a series of high-level meetings in San Francisco between Huang and the chiefs of Korean tech giants, including Samsung Electronics Chairman Lee Jae-yong and Hyundai Motor Group Executive Chair Euisun Chung.
The San Francisco AI Declaration
These corporate deals coincided with a broader geopolitical push by South Korean President Lee Jae Myung. During his San Francisco AI Declaration
on July 24, Lee outlined a vision to make Korea an indispensable pillar of the global AI supply chain.
The president’s strategy involves a multihub semiconductor production network spanning regional provinces and the Seoul metropolitan area. This domestic supply chain is intended to act as a national safeguard to ensure production remains uninterrupted during crises and to prevent future global chip shortages.
The financial scale of these ambitions is immense.
SK Hynix’s Nasdaq Debut and Capital Strategy
To fund these massive infrastructure developments, SK Hynix listed on the Nasdaq on July 10, 2026. The listing, which raised approximately $26.5 billion, is one of the largest share sales in market history.
This capital injection is critical because new HBM supply is not expected to arrive before late 2027, according to analysis reported by 247WallSt. Until then, the memory shortage will likely keep prices at record highs, further cementing the advantage of any company—like Nvidia—that has already locked in its supply.
Worth a look
