Samsung Ships 12-Layer HBM4E Samples as Shares Jump 6.5%
Updated
Updated · CNBC · May 29
Samsung Ships 12-Layer HBM4E Samples as Shares Jump 6.5%
6 articles · Updated · CNBC · May 29
Samsung said it has started shipping samples of its 12-layer HBM4E chip to customers globally, sending the stock up as much as 6.51% before it traded 3.67% higher at 310,500 won.
The new AI memory chip is a 48GB part with speeds of up to 16 Gbps, and Samsung said it improves energy efficiency and thermal performance while lifting capacity by more than 30% from the prior generation.
HBM chips are critical for AI accelerators such as Nvidia's Rubin and Google's Ironwood TPU, giving Samsung a fresh product push as demand rises for faster data handling in advanced systems.
Samsung began shipping HBM4 chips in February and plans to add 8-layer 32GB and 16-layer 64GB HBM4E versions, part of its effort to close ground on SK Hynix in next-generation AI memory.
With SK Hynix using TSMC's advanced nodes, can Samsung's in-house process win the next-gen AI memory race?
With shortages predicted into 2027, will Samsung's new chip ease the AI bottleneck or just intensify the supply war?
Samsung’s HBM4E Leap: Early Shipments, Strategic Partnerships, and the High-Stakes AI Memory Market of 2026
Overview
In May 2026, Samsung shipped samples of its 12-layer HBM4E memory, marking a major milestone in the high-bandwidth memory market. This achievement puts Samsung ahead of competitors and at the forefront of next-generation AI system development. The company’s rapid progress is driven by its integrated semiconductor portfolio, which includes memory, foundry, logic design, and advanced packaging. By using the same core and base die combination as its previous HBM4 products, Samsung can iterate efficiently and bring products to market faster. This approach ensures a stable supply of advanced memory for the fast-growing artificial intelligence sector.