Micron Technology shares dropped 3.2% Monday as Samsung Electronics accelerated its next-generation memory chip timeline. The South Korean manufacturer will begin mass production of HBM4 chips this month, according to industry sources cited by Yonhap News Agency.
Samsung's new memory targets Nvidia's upcoming Vera Rubin AI accelerator platform, with the company reportedly passing Nvidia's quality certification process and securing purchase orders. Shipments are scheduled to start after the Lunar New Year holiday, and Samsung shares surged 6.4% on the announcement.
The HBM4 chips deliver data transfer speeds of 11.7 gigabits per second, a 22% improvement over previous HBM3E generation standards. Memory bandwidth reaches 3 terabytes per second per stack, with 12-high stacking enabling 36 gigabyte capacities. Future 16-high configurations could expand to 48 gigabytes.
Samsung's production ramp comes as memory prices surge across the industry. TrendForce's latest February 2, 2026 forecast shows conventional DRAM contract prices rising 90-95% in the first quarter of 2026, up from its earlier January 5 estimate of 55-60%. NAND flash prices are now expected to increase 55-60% during the same period, up from the earlier 33-38% estimate.
The HBM market currently shows SK hynix leading with 62% share, followed by Micron at 21% and Samsung at 17%. Counterpoint Research analysts expect Samsung's position to strengthen as HBM4 enters full-scale supply this year. Market tracker SemiAnalysis projects SK hynix capturing 70% of the HBM4 market, with Samsung taking the remaining 30%.
Samsung Electronics reported record fourth-quarter operating profit of 20 trillion won ($13.8 billion), marking a 200% year-over-year increase. The company expects its HBM sales volume to triple this year compared to 2025 levels.
Additional production lines at Samsung's Pyeongtaek Campus Line 4 are expected to produce roughly 100,000 to 120,000 wafers monthly dedicated to 1c DRAM used in HBM4 products.
Memory chip manufacturers face tightening supply conditions as AI infrastructure demands accelerate. Nvidia's CEO has warned that semiconductor production capacity may need to double to meet explosive AI demand.
Samsung's accelerated timeline represents a strategic recovery for the company, which trailed competitors in earlier high-bandwidth memory generations. Nvidia will reportedly unveil Vera Rubin accelerators incorporating Samsung's HBM4 at its GTC 2026 conference in March 2026 (March 16-19), with timing coordinated with Samsung's production schedule.
