Samsung Electronics announced that it has begun shipping what it describes as the industry’s first 12-layer HBM4E (High Bandwidth Memory 4E) samples to major global customers, expanding its next-generation AI memory portfolio and strengthening its position in the rapidly growing AI infrastructure market.
The announcement follows Samsung’s mass production and commercial shipment of HBM4 earlier in 2026. With HBM4E, the company is targeting the increasing performance requirements of artificial intelligence workloads, hyperscale data centers, and large language model deployments.
Samsung said the new HBM4E memory delivers a stable pin speed of 14 gigabits per second (Gbps), with performance scalable up to 16Gbps. The company noted that this represents more than a 20% improvement over HBM4. The memory also provides bandwidth of up to 3.6 terabytes per second (TB/s) per stack, helping support intensive AI computing tasks and next-generation AI systems.
The initial HBM4E product features a 12-layer design with a capacity of 48GB, which Samsung said is more than a 30% increase compared to the previous generation. The company also plans to expand the lineup with 32GB (8-layer) and 64GB (16-layer) versions to meet varying customer requirements.
Samsung attributed the HBM4E’s performance gains to technologies refined through its HBM4 production experience. The memory utilizes the company’s sixth-generation 10-nanometer-class DRAM process technology, known as 1c, along with a 4nm logic base die produced by Samsung Foundry. According to Samsung, this combination enhances manufacturing stability, yield, and overall performance.
The company also highlighted improvements in power efficiency and thermal management. Through advanced low-power design technologies and packaging optimizations, Samsung said HBM4E improves energy efficiency by 16% and enhances thermal resistance characteristics by more than 14% compared to the previous generation. These improvements are designed to help AI data centers manage increasingly demanding workloads while reducing energy consumption and improving long-term reliability.
Samsung plans to begin mass production of HBM4E in alignment with customer deployment schedules following sample evaluation and optimization processes.
The company noted that customer feedback on its HBM4 products, introduced earlier this year, has been positive, particularly regarding performance and energy efficiency. Samsung’s HBM4 became the first HBM4 product in the industry to enter mass production and achieved speeds of 11.7Gbps during system-in-package testing.
As AI adoption continues to accelerate worldwide, Samsung said its expanding HBM portfolio, combined with its capabilities across memory, foundry services, logic design, and advanced packaging, positions the company to support growing semiconductor demand for AI infrastructure.
KEY QUOTE:
“Following the successful mass production of HBM4, Samsung has once again demonstrated its distinct technological edge with HBM4E. Through our advanced manufacturing capabilities and preemptive infrastructure investments, we will continue to drive the growth of the global AI memory market.”
Sang Joon Hwang, Executive Vice President And Head Of Memory Development, Samsung Electronics
