Samsung and SK hynix stand to gain as Nvidia locks in memory supply
The U.S. tech powerhouse has more than doubled its supply and capacity commitments to $279 billion, a move that is likely to benefit its top chip suppliers.
Nvidia founder and CEO Jensen Huang listens as U.S. President Donald Trump speaks in the Oval Office of the White House about Dulles International Airport modernization in Washington on July 29.AP/YONHAP
Nvidia said Wednesday it more than doubled its supply and capacity commitments to $279 billion last quarter, mostly to lock in memory and manufacturing capacity — a signal of stronger demand ahead for Samsung Electronics and SK hynix.
The commitments, up from $119 billion three months earlier, cover the data center infrastructure Nvidia is building, primarily the memory and manufacturing capacity it will need over the coming years.
Nvidia reported revenue of $96.2 billion for its fiscal second quarter, which ended July 26, up 106 percent from a year earlier. Data center revenue rose 117 percent to $89 billion.
"We've partnered with our extensive network of suppliers to secure the critical components needed to meet demand for the next several years," Nvidia chief financial officer Colette Kress said in written commentary alongside the results. "Our commitments increased from $119 billion last quarter to $279 billion, primarily related to the procurement of memory."
Kress also told analysts on the earnings call that Nvidia is working closely with all three of its major memory suppliers — Samsung, SK hynix and Micron — to expand capacity.
Samsung and SK hynix make most of that memory. High bandwidth memory (HBM) stacks dynamic random-access memory (DRAM) chips vertically to move large volumes of data quickly, and it is the memory that goes into the accelerators running AI workloads.
Competition is sharpening around HBM4, the sixth generation, as Nvidia's next platform reaches volume. Vera Rubin entered full production and pairs Nvidia's Vera processors with its Rubin graphics chips and large amounts of HBM4.
Nvidia CEO Jensen Huang, left, receives an SK hynix HBM4 semiconductor wafer as a gift from SK Group Chairman and Korea Chamber of Commerce and Industry Chairman Chey Tae-won during the APEC CEO Summit at the Gyeongju Arts Center in Gyeongju, North Gyeongsang, on Oct. 31, 2025.KIM SEONG-RYONG
SK hynix leads the HBM market and has been Nvidia's main supplier across earlier generations. Nvidia says the AI infrastructure SK Group is building will run on Vera Rubin systems carrying SK hynix HBM4. Samsung has already moved. It began mass production of HBM4 in February and said in March that the chips are designed for Vera Rubin.
Memory prices could yet slow the buildout. Server DRAM prices rose 64 percent in the second half of last year, and the research firm TrendForce expects a 260 percent jump this year.
"If rising memory prices push up the cost of building AI infrastructure, it could become difficult for Big Tech companies to maintain the current pace of investment," said Shin Jung-ho, the head of the research center at LS Securities.
"If those costs are passed on to customers, they could also weigh on the profitability of AI data center investment."
This article was originally written in Korean and translated by a bilingual reporter with the help of generative AI tools. It was then edited by a native English-speaking editor. All AI-assisted translations are reviewed and refined by our newsroom.
