Nvidia CEO Jensen Huang visited the SK Hynix booth during Computex 2026 to publicly request increased production of high-bandwidth memory [1].
The interaction highlights the critical dependency of AI chipmakers on specialized memory suppliers to meet the soaring demand for artificial intelligence infrastructure. As Nvidia continues to scale its GPU offerings, the availability of high-bandwidth memory (HBM) remains a primary bottleneck for industry growth.
During the afternoon of the second day of the exhibition [1], Huang made a surprise appearance at the booth. He met with SK Group Chairman Choi Tae-won, SK Hynix President Kwak No-jeong, and AI-Infrastructure head Kim Joo-sun [1]. The meeting served as a public display of goodwill and a reinforcement of the strategic ties between the two companies.
While at the booth, Huang signed an HBM4E wafer with the phrase "Please make more" [1]. He also wrote "(HBM을) 더 만들어달라" on the wafer [1]. The request was a direct appeal for the company to accelerate the manufacturing of these essential components.
Further demonstrating his support, Huang signed a 192 GB socam [1]. On this component, he wrote "소캠 사랑해 (LOVE SOCAMM)" [1]. Following these gestures, Huang took a group photo with the SK executives to commemorate the visit.
Computex 2026 is recognized as the largest IT exhibition in Asia [1]. The event typically serves as a venue for major hardware announcements and strategic partnerships. Huang's decision to visit the booth unexpectedly underscores the urgency of the memory supply chain in the current AI race.
“"Please make more"”
The public nature of this request indicates that the demand for HBM4E and high-capacity socams is outpacing current supply capabilities. By personally appealing to SK Hynix leadership, Nvidia is signaling to the market that memory availability is a top priority for its 2026 hardware roadmap, potentially pressuring other suppliers to increase their own output to remain competitive.
