The two companies at the heart of the artificial intelligence hardware boom are tightening their alliance, and the message coming out of Seoul this week is one that will shape technology markets for years. Nvidia and South Korea’s SK Group are preparing to detail a cooperation plan, even as Nvidia chief executive Jensen Huang warns that the shortage of advanced memory chips powering AI systems will persist for quite a few years. According to CNBC, SK hynix group Chairman Chey Tae-won and Huang planned to brief the media together, capping a flurry of high level meetings between the American chip designer and the Korean conglomerates that supply its most critical components.
For investors, technologists, and anyone tracking the economics of artificial intelligence, the development carries weight well beyond a single partnership announcement. It confirms that the bottleneck in AI computing has shifted, at least for now, from the graphics processors that perform calculations to the high-bandwidth memory that feeds them data. And it signals that the company best positioned to relieve that bottleneck, SK hynix, has become indispensable to Nvidia’s roadmap.
The Memory at the Center of the AI Boom
To understand why a memory chip supplier commands this much attention, it helps to understand what high-bandwidth memory, or HBM, actually does. Modern AI accelerators perform staggering volumes of calculations, but those calculations are useless unless data can be delivered to the processor fast enough to keep it busy. HBM is a specialized form of memory that stacks chips vertically and connects them with dense, high-speed pathways, delivering far more bandwidth than conventional memory designs. In an AI data center, HBM is the difference between a processor running at full capacity and one starved for data.
SK hynix has emerged as the dominant force in this market. The company supplies Nvidia with HBM4, its latest generation of high-bandwidth memory, along with high-performance, low-power products such as LPDDR5X. That supply relationship has propelled SK hynix to extraordinary heights, including a market capitalization that reached $1 trillion, a milestone the company and Nvidia executives met in Taipei to celebrate earlier in the year. The achievement underscores how the AI buildout has minted enormous value for the firms that supply its essential ingredients, a dynamic explored in our coverage of SK hynix crossing the $1 trillion valuation mark.
The strategic importance of memory was captured in a single, telling gesture. During a visit to SK hynix’s booth at the Computex trade show, Huang reviewed the company’s next-generation memory products and left handwritten notes on display wafers. On an HBM4E sample, he wrote “Please make more.” On a SOCAMM memory module, he wrote “Love SOCAMM.” The phrase “please make more” is half joke and half desperate plea, a public acknowledgment from the most powerful figure in AI hardware that he cannot get enough of what SK hynix produces.
A Shortage That Is Not Going Away
The most consequential takeaway from this week is Huang’s blunt assessment that the memory shortage will not resolve quickly. His warning that constraints could persist for quite a few years reframes the supply situation from a temporary imbalance into a structural feature of the AI era. Demand for AI compute is growing faster than the industry can build the fabrication capacity to produce the advanced memory that compute requires, and closing that gap takes years because new memory fabs are enormously expensive and slow to construct.
For the companies that make HBM, this is a remarkable position to occupy. A prolonged shortage means sustained pricing power, full order books, and the ability to command premium terms from even the largest customers. It explains why SK hynix, Samsung, and Micron have all raced to qualify their HBM4 products for Nvidia’s systems, and why securing allocation has become a strategic priority for every buyer of AI hardware.
The qualification race reached a key milestone recently. Huang publicly validated that all three leading memory manufacturers, Samsung, SK hynix, and Micron, had completed the qualification process to supply HBM4 for Nvidia’s Vera Rubin platform, the company’s next-generation AI computing system that has now entered full production. Industry analysts estimate that SK hynix holds the largest share of HBM4 volume for Vera Rubin systems, at roughly 60 to 70 percent, with Samsung at approximately 25 to 30 percent and Micron supplying the remainder. That allocation reflects SK hynix’s technological lead and the depth of its partnership with Nvidia.
Beyond Memory: Robotics, Physical AI, and a Korean Bet
The cooperation taking shape between Nvidia and SK extends beyond the immediate question of memory supply. Huang’s trip to South Korea has been packed with meetings that suggest a broader strategic vision. He has signaled interest in robotics as an area for investment in the country, and business observers expect the talks to move beyond AI semiconductor cooperation to include robotics and physical AI, the emerging field focused on integrating artificial intelligence with real-world machines and systems.
This points to where Nvidia sees the next frontier. Having established dominance in the data center, the company is looking toward a future in which AI controls physical systems, from autonomous machines to industrial robots, and South Korea’s deep industrial base in manufacturing, electronics, and automotive production makes it a natural partner for that ambition. Huang’s itinerary reportedly included a visit to Seoul National University to meet AI researchers and tour its robotics research center, alongside a closed-door meeting with local robotics and AI startups.
The relationship building has been intense and personal. Huang has met SK Group Chairman Chey at least three times in 2026 alone. The pattern of high level engagement, including a Korea Partner Night gathering in Taipei attended by senior executives from Samsung, SK hynix, LG Electronics, Hyundai Motor Group, and Naver, reflects how central the Korean technology ecosystem has become to Nvidia’s plans. These are not transactional supplier meetings. They are the cultivation of a strategic alliance spanning multiple industries.
What It Means for Markets and the AI Trade
The implications for investors are significant. A structural memory shortage that persists for years changes the calculus for the entire AI supply chain. Memory makers become long-term beneficiaries of sustained demand and pricing power, while AI system builders must factor component scarcity into their growth plans. The companies that control scarce, essential inputs capture an outsized share of the value created by the AI boom, a theme that runs through our analysis of the best AI stocks to watch in 2026.
The dynamic also reinforces a broader lesson about where profit accumulates in a technology gold rush. Just as the most durable value in past booms often flowed to the suppliers of essential infrastructure rather than the most visible end products, the AI era is rewarding the firms that solve its physical bottlenecks. Nvidia itself has internalized this logic, investing heavily across its supply chain to secure the inputs its roadmap depends on, an approach mirrored in its multibillion-dollar bet on silicon photonics to overcome the limits of copper interconnects.
For SK hynix and its Korean peers, the message from this week is overwhelmingly positive. They sit at the center of the most important supply relationship in technology, supplying a component that Nvidia’s chief executive has publicly begged them to produce more of, in a market where shortage is expected to persist for years. That is about as favorable a competitive position as a manufacturer can occupy.
The deepening Nvidia-SK alliance, set against Huang’s warning of a prolonged shortage, captures the defining tension of the current AI moment. Demand is effectively unlimited, supply is constrained by physics and capital, and the companies that can expand capacity for the scarcest components hold the keys to how fast the AI future arrives. This week in Seoul, those companies signaled that they intend to build that future together.
