Samsung Electronics is preparing to commence mass production of its sixth-generation high-bandwidth memory chips at its Pyeongtaek campus, with initial shipments to Nvidia and AMD expected as early as the third week of February. The milestone, reported by multiple Korean industry sources and corroborated by TrendForce, would make Samsung the first company globally to deliver HBM4 at commercial scale, a distinction that carries considerable competitive significance in a market where SK Hynix has dominated for the past two years.
The timing is deliberate. Samsung’s HBM4 chips are destined for Nvidia’s Vera Rubin AI accelerator platform, which is scheduled for launch in the second half of 2026 and is expected to be showcased at Nvidia’s GTC conference in March. By beginning shipments ahead of SK Hynix, which has reportedly delayed its own HBM4 mass production from February to March or April, Samsung positions itself to capture a larger share of initial Vera Rubin production volumes. The company cleared Nvidia’s final qualification tests in recent weeks, receiving purchase orders rather than additional sample requests, a progression that indicates the product has moved beyond evaluation into confirmed supply.
The technical specifications reinforce Samsung’s competitive claim. Its HBM4 achieves a data transfer rate of 11.7 gigabits per second, exceeding the 10 Gbps requirement set by both Nvidia and AMD by a substantial margin. Memory bandwidth per stack reaches approximately 3 terabytes per second, roughly 2.4 times the bandwidth of the preceding HBM3E generation. Samsung has attributed this performance advantage to its use of sixth-generation 1c DRAM on a 10-nanometer class process, one generation ahead of competitors, combined with a 4-nanometer foundry process for the logic die that controls the memory stack. The vertically integrated production model, spanning DRAM fabrication, logic die manufacturing, and advanced packaging within Samsung’s own facilities, gives the company a structural advantage in lead time over rivals that depend on external foundries such as TSMC for their logic components.
For investors, the significance of Samsung’s HBM4 progress extends beyond the immediate product cycle. The company’s semiconductor division issued a rare public apology in October 2025 for falling behind on HBM3E deliveries to Nvidia, an episode that damaged Samsung’s credibility and contributed to SK Hynix capturing a dominant market share estimated at 53% to 61% depending on the research firm. Samsung’s ability to ship HBM4 first would represent a tangible reversal of that narrative. Morgan Stanley analysts have projected that Samsung’s memory profits could rise more than 300% in 2026, reflecting both higher margins from AI-driven chip demand and the recovery of market share in the most valuable segment of the memory industry.
The competitive landscape remains intense. SK Hynix, which completed development of HBM4 in September 2025 and shipped samples to customers months before Samsung, has not conceded the race. The company’s decision to delay mass production appears linked to adjustments in Nvidia’s product strategy rather than technical shortcomings. SK Hynix continues to supply HBM3E as its primary product and has secured what industry sources describe as the majority of Nvidia’s initial HBM4 orders. Samsung is reportedly negotiating to supply more than 30% of Nvidia’s HBM4 requirements for 2026, a meaningful share but still a minority position relative to SK Hynix’s allocation.
Samsung’s broader capacity expansion plans add context to the production ramp. The company aims to increase total HBM production capacity by approximately 50% by year-end 2026, targeting 250,000 wafers per month, up from a current base of around 170,000. That expansion involves converting existing lines and scaling up the Pyeongtaek 4 facility. Samsung has also signaled plans to expand its next-generation DRAM production lines, increasing capacity for the advanced memory used in HBM4 by as much as 70%. These capital commitments reflect a company that views the AI memory cycle as structural rather than cyclical and is willing to invest aggressively to reclaim its historical position as the dominant force in global memory production.
The market implications are considerable. Samsung Electronics shares have been among the strongest performers on the KOSPI in early 2026, and confirmation of commercial HBM4 shipments would provide an additional catalyst. For portfolio managers with exposure to the Korean semiconductor sector, the Samsung-SK Hynix dynamic has become the defining competitive axis. Both companies are investing tens of billions of dollars in capacity expansion, both are targeting the same hyperscale data center customers, and both are leveraging AI-driven demand to support pricing power across their broader memory product lines. The outcome of the HBM4 production race will shape relative market share, margin trajectories, and investor sentiment for the remainder of the year.
