HBM4 Volume and SK Hynix’s Dominance

Huang used the visit to settle months of supply-chain speculation around HBM4. Stepping off his plane at Seoul’s Gimpo airport on June 5, he confirmed that all three major memory manufacturers—Samsung, SK Hynix, and Micron—have been certified and are in active production for Nvidia’s Vera Rubin HBM4 slots . The first systems are scheduled to ship in Q3 2026 .

Industry analysts estimate SK Hynix holds roughly 60–70% of the HBM4 volume allocated to Vera Rubin, with Samsung at approximately 25–30% and Micron supplying the remainder . Nvidia has not disclosed official allocation figures, and earlier in 2026 analyst firm SemiAnalysis had projected Micron’s share at 0%—a question the June 5 certification of all three suppliers appeared to settle definitively .

The Broader Partnership: Beyond Memory

Huang’s Seoul schedule culminated in a Monday-morning tea meeting on June 8 with SK Group Chairman Chey Tae-won at the SK Seorin Building in Jongno, their fifth gathering of the week . SK Hynix confirmed that Chey and Huang planned to brief the media together afterward . The expected collaboration plan covers procurement and supply of high-performance memory for AI, with Huang signaling that the deals will be “even larger in the second half and next year” .

The partnership blueprint likely includes:

• Deeper HBM and DRAM supply commitments
• Co-investment in production capacity
• Expanded R&D cooperation
• Involvement from SK Telecom, which attended the Sunday dinner alongside SK Hynix executives

Huang framed the alliance expansively, stating Nvidia is “growing together with Hyundai Motor, LG, SK Hynix, Samsung, and Naver” across the full AI stack . He brought what he called a “big gift” to Korea: four new businesses—Vera Rubin, Vera, RTX Spark, and Jetson Thor—along with plans for a prominent Nvidia research center in the country .

Robotics, GTC, and the Next Sector

Huang declared robotics as South Korea’s “next major sector” and promised “surprises” for Korea . He highlighted that Korea is an “indispensable core axis” of Nvidia’s AI ecosystem and expressed intent to expand collaboration beyond semiconductors into robotics and AI factories . In a notable aside, Huang suggested Seoul could host Nvidia’s annual GTC developer conference in the future—a signal of how central the Korean market has become to Nvidia’s physical AI ambitions .

The Memory Shortage and SK Hynix’s Financial Position

Away from the fried-chicken diplomacy, Huang issued a stark warning: the memory shortage will last “quite a few years” as AI demand far outpaces supply . SK Hynix’s HBM capacity is already sold out through 2027, and some customers are offering advance payments of 10–30% to secure capacity . Against this backdrop, the company’s financials are remarkable:

• The stock has nearly tripled since the start of 2026, trading at roughly 1.97 million won
• It has surged more than 60% in the past 30 days alone
• Over one year, it is up 207%, trading at a P/E of 10.8× compared to many peers above 15×
• Analyst consensus is a “Strong Buy”

To fund factory expansion in Yong-in, M15X, and a packaging facility in Indiana, SK Hynix has confidentially filed a Form F-1 with the SEC for a U.S. ADR listing targeting the second half of 2026, expected to raise an estimated $10 billion to $14 billion . Chairman Chey first confirmed the listing review at GTC 2026 in March, calling it a move to make SK Hynix “a more global company,” and the company has since told investors it received “tremendously positive” feedback on the plan .

With the Vera Rubin platform in full production, the Vera CPU locked to SK Hynix DRAM, and a U.S. listing on the horizon, Huang’s Seoul trip marked the moment Nvidia’s AI supply chain went from transactional to structural—anchored in Korean manufacturing and capital markets.

HBM4 Volume and SK Hynix’s Dominance

Huang used the visit to settle months of supply-chain speculation around HBM4. Stepping off his plane at Seoul’s Gimpo airport on June 5, he confirmed that all three major memory manufacturers—Samsung, SK Hynix, and Micron—have been certified and are in active production for Nvidia’s Vera Rubin HBM4 slots . The first systems are scheduled to ship in Q3 2026 .

Industry analysts estimate SK Hynix holds roughly 60–70% of the HBM4 volume allocated to Vera Rubin, with Samsung at approximately 25–30% and Micron supplying the remainder . Nvidia has not disclosed official allocation figures, and earlier in 2026 analyst firm SemiAnalysis had projected Micron’s share at 0%—a question the June 5 certification of all three suppliers appeared to settle definitively .

The Broader Partnership: Beyond Memory

Huang’s Seoul schedule culminated in a Monday-morning tea meeting on June 8 with SK Group Chairman Chey Tae-won at the SK Seorin Building in Jongno, their fifth gathering of the week . SK Hynix confirmed that Chey and Huang planned to brief the media together afterward . The expected collaboration plan covers procurement and supply of high-performance memory for AI, with Huang signaling that the deals will be “even larger in the second half and next year” .

The partnership blueprint likely includes:

• Deeper HBM and DRAM supply commitments
• Co-investment in production capacity
• Expanded R&D cooperation
• Involvement from SK Telecom, which attended the Sunday dinner alongside SK Hynix executives

Huang framed the alliance expansively, stating Nvidia is “growing together with Hyundai Motor, LG, SK Hynix, Samsung, and Naver” across the full AI stack . He brought what he called a “big gift” to Korea: four new businesses—Vera Rubin, Vera, RTX Spark, and Jetson Thor—along with plans for a prominent Nvidia research center in the country .

Robotics, GTC, and the Next Sector

Huang declared robotics as South Korea’s “next major sector” and promised “surprises” for Korea . He highlighted that Korea is an “indispensable core axis” of Nvidia’s AI ecosystem and expressed intent to expand collaboration beyond semiconductors into robotics and AI factories . In a notable aside, Huang suggested Seoul could host Nvidia’s annual GTC developer conference in the future—a signal of how central the Korean market has become to Nvidia’s physical AI ambitions .

The Memory Shortage and SK Hynix’s Financial Position

Away from the fried-chicken diplomacy, Huang issued a stark warning: the memory shortage will last “quite a few years” as AI demand far outpaces supply . SK Hynix’s HBM capacity is already sold out through 2027, and some customers are offering advance payments of 10–30% to secure capacity . Against this backdrop, the company’s financials are remarkable:

• The stock has nearly tripled since the start of 2026, trading at roughly 1.97 million won
• It has surged more than 60% in the past 30 days alone
• Over one year, it is up 207%, trading at a P/E of 10.8× compared to many peers above 15×
• Analyst consensus is a “Strong Buy”

To fund factory expansion in Yong-in, M15X, and a packaging facility in Indiana, SK Hynix has confidentially filed a Form F-1 with the SEC for a U.S. ADR listing targeting the second half of 2026, expected to raise an estimated $10 billion to $14 billion . Chairman Chey first confirmed the listing review at GTC 2026 in March, calling it a move to make SK Hynix “a more global company,” and the company has since told investors it received “tremendously positive” feedback on the plan .

With the Vera Rubin platform in full production, the Vera CPU locked to SK Hynix DRAM, and a U.S. listing on the horizon, Huang’s Seoul trip marked the moment Nvidia’s AI supply chain went from transactional to structural—anchored in Korean manufacturing and capital markets.