The market responded sharply. SK Hynix's ADRs surged 27.29% on July 14, their third trading day, as perceptions of undervaluation spread .

This praise followed a pattern. In June 2026, Huang had pleaded with SK Hynix to "please make more" memory chips, warning that the AI memory shortage would persist for "the next several years" . At Computex 2026, he wrote "Please Make More" on an HBM4E wafer at SK Hynix's booth . Huang has repeatedly emphasized that memory demand will exceed capacity, stating that even if the supply chain quadruples every year, "the next ten years won't be enough for AI" .

Japan AI Collaboration: 'The Starting Point of Japan's AI'

On July 15, Huang announced that the following day (July 16) he would unveil a large-scale AI partnership with the Japanese government, along with multifaceted cooperation with major Japanese companies in AI and robotics .

He framed the moment dramatically: "Tomorrow can be called the starting point of Japan's AI" .

Nvidia's official blog confirmed a sweeping "full-stack AI and robotics" ecosystem push across Japan , including:

• Healthcare & life sciences — AI drug discovery consortia (Tokyo-1 with Astellas, Daiichi Sankyo, Ono Pharmaceuticals, and Eisai using NVIDIA BioNeMo); surgical robots from Kawasaki Heavy Industries; next-generation CT systems from Canon and Fujifilm powered by NVIDIA GPUs.
• Industrial AI & robotics — Partnerships with Japanese manufacturing leaders on physical AI, vision AI agents, and automation.

The trip also had a diplomatic backstory. Huang had previously sparked a "Japan passing" controversy by visiting South Korea and Taiwan twice without stopping in Japan, stirring anxiety in Tokyo's IT industry . This visit was seen as a correction, timed with the 30th anniversary of Nvidia's partnership with SEGA .

Why HBM Supply Matters for the AI Race

The two announcements are connected by a deeper theme: the AI chip supply chain is under structural strain, and Nvidia is actively reinforcing its partnerships with the suppliers—SK Hynix for memory and Japan for full-stack AI infrastructure.

SK Hynix currently holds an effective monopoly on HBM3E supply, which is integrated into Nvidia's flagship Blackwell AI chip . Huang has warned that the memory shortage is not cyclical but structural, with wafer supply running more than 20% below demand and the gap expected to persist through 2030 . SK Hynix plans to double DRAM wafer capacity by 2030, but Huang has publicly stated that expansion is insufficient .

The Japan partnership, meanwhile, aligns with Huang's long-standing message: "You must build it yourself because it's your intelligence" . He has urged Japan to generate more power and build out AI infrastructure, positioning the country as a natural leader in "physical AI" and robotics due to its industrial manufacturing base .

The market responded sharply. SK Hynix's ADRs surged 27.29% on July 14, their third trading day, as perceptions of undervaluation spread .

This praise followed a pattern. In June 2026, Huang had pleaded with SK Hynix to "please make more" memory chips, warning that the AI memory shortage would persist for "the next several years" . At Computex 2026, he wrote "Please Make More" on an HBM4E wafer at SK Hynix's booth . Huang has repeatedly emphasized that memory demand will exceed capacity, stating that even if the supply chain quadruples every year, "the next ten years won't be enough for AI" .

Japan AI Collaboration: 'The Starting Point of Japan's AI'

On July 15, Huang announced that the following day (July 16) he would unveil a large-scale AI partnership with the Japanese government, along with multifaceted cooperation with major Japanese companies in AI and robotics .

He framed the moment dramatically: "Tomorrow can be called the starting point of Japan's AI" .

Nvidia's official blog confirmed a sweeping "full-stack AI and robotics" ecosystem push across Japan , including:

• Healthcare & life sciences — AI drug discovery consortia (Tokyo-1 with Astellas, Daiichi Sankyo, Ono Pharmaceuticals, and Eisai using NVIDIA BioNeMo); surgical robots from Kawasaki Heavy Industries; next-generation CT systems from Canon and Fujifilm powered by NVIDIA GPUs.
• Industrial AI & robotics — Partnerships with Japanese manufacturing leaders on physical AI, vision AI agents, and automation.

The trip also had a diplomatic backstory. Huang had previously sparked a "Japan passing" controversy by visiting South Korea and Taiwan twice without stopping in Japan, stirring anxiety in Tokyo's IT industry . This visit was seen as a correction, timed with the 30th anniversary of Nvidia's partnership with SEGA .

Why HBM Supply Matters for the AI Race

The two announcements are connected by a deeper theme: the AI chip supply chain is under structural strain, and Nvidia is actively reinforcing its partnerships with the suppliers—SK Hynix for memory and Japan for full-stack AI infrastructure.

SK Hynix currently holds an effective monopoly on HBM3E supply, which is integrated into Nvidia's flagship Blackwell AI chip . Huang has warned that the memory shortage is not cyclical but structural, with wafer supply running more than 20% below demand and the gap expected to persist through 2030 . SK Hynix plans to double DRAM wafer capacity by 2030, but Huang has publicly stated that expansion is insufficient .

The Japan partnership, meanwhile, aligns with Huang's long-standing message: "You must build it yourself because it's your intelligence" . He has urged Japan to generate more power and build out AI infrastructure, positioning the country as a natural leader in "physical AI" and robotics due to its industrial manufacturing base .