Taipei, Taiwan, June 1, 2026 (Globe Newswire) — Micron Technology, Inc. (Nasdaq: MU) today announced during COMPUTEX 2026 a showcase of its complete portfolio of AI-optimized memory and storage solutions powering next-generation AI data centers and intelligent edge applications. As AI workloads expand from training to large-scale inference, including inference-focused and agent-based systems, demands on memory and storage are increasing at every layer of the compute stack and memory hierarchy.
“The length of the AI context is increasing 30x per year;
1
Memory content per server has doubled over the past three years;
2
said Sumit Sadhana, Vice President and Chief Business Officer of Micron Technology. “System performance is now more dependent on memory bandwidth and memory capacity than ever before.” This structural shift in the semiconductor ecosystem has made memory and storage essential strategic assets, and Micron is leading with a range of industry firsts and industry best products, from HBM to DRAM and NAND solutions designed for the AI era. ”
Memory and storage as the foundation of AI data center performance
Micron’s data center memory and storage portfolio is purpose-built to address every layer of the AI infrastructure hierarchy. High-bandwidth memory (HBM) powers fast model execution and hot key value (KV) caching, while LPDDR and DDR provide system memory for orchestration and long context expansion (LPDDR is more power efficient). Data center SSDs complement the stack, providing high-performance drives for persistent KV caching needs and high-capacity drives for large data lakes. This layered memory architecture, with Micron at the center of each layer, optimizes latency, bandwidth, power, capacity, and cost, offloading GPUs and maximizing data center token generation. The latest milestones across Micron’s portfolio demonstrate this momentum.
-
HBM:
With HBM4 36GB 12H, each 2x increase in bandwidth increases Large Language Model (LLM) inference throughput, measured in tokens per second, by 2.6x.
3
-
Sokam:
Micron is the only provider of the world’s highest capacity 256GB SOCAMM2, extending its leadership in low-power data center memory by delivering one-third the power and one-third the footprint compared to standard RDIMMs.
4
-
Large capacity RDIMM:
Micron has sampled its cutting-edge 1γ (1 gamma) technology with a 256GB DDR5 RDIMM. It is capable of speeds of up to 9,200 megatransfers per second (MT/s), 40% faster than modules currently in production, and uses more than 40% lower operating power compared to two 128GB modules.
5
-
Data center SSD:
Micron 9650 SSD was the world’s first commercially available PCIe
®
Gen6 SSDs are designed to deliver high performance for AI inference and training workloads. Now available in up to 245 TB, Micron 6600 ION sets new benchmarks in density and power efficiency, reducing rack footprint by 82%
6
Cut power consumption in half
7
compared to HDD-based deployments.
AI starts in the cloud. Edge delivers that value
As AI inference expands from data centers to PCs, smartphones, vehicles, and embedded systems, the demands on memory and storage are fundamentally changing. Micron is engineering for this change. While high-density DRAM keeps AI models and agents resident and running, Micron’s storage solutions evolve into an active work layer, supporting everything from local model caching on the AI PC to real-time sensor fusion in the vehicle, delivering faster, more responsive AI experiences at every edge.
-
LPCAM:
LPCAMM2 is a modular, low-power 128-bit dual channel design for thin and light PCs, delivering up to 9,600 MT/s with LPDDR5X. -
GDDR:
GDDR7 delivers up to 1.5 TB/s of system bandwidth, 60% higher than GDDR6
8
Increase AI inference throughput by up to 33%.
9
-
LPDDR:
LPDDR5X delivers industry-leading low-power performance for real-time AI processing across PCs, smartphones, robotics, and next-generation automotive platforms. -
Client SSD:
Micron 4600 PCIe Gen5 NVMe™ SSD loads LLM in less than 1 second.
10
107% more energy efficient than previous generation Gen4 SSDs.
11
-
Automotive UFS:
UFS 4.1 delivers up to 4.2 GB/s, double the previous generation, 115°C thermal protection and functional safety compliance for advanced driver assistance systems (ADAS) and in-vehicle AI.
A shared commitment to the future of AI
AI fundamentally recasts memory as a defining strategic asset, requiring memory and compute to be designed together for optimal results. Building on our multi-generational leadership in DRAM and NAND process technologies, most recently 1γ DRAM and G9 NAND, Micron is deepening technology collaboration with partners across the ecosystem through collaborative design and engineering to bring AI platforms to market faster and with better system-level optimization. Backed by significant manufacturing investments across the United States, India, Japan, Singapore and Taiwan, Micron is positioned to deliver these innovations at scale.
COMPUTEX showcase
During COMPUTEX 2026, Micron will host an invitation-only product showcase at its Taipei office in TFC Plaza from June 2nd to June 4th. Contact your Micron representative to schedule a visit.
About Micron Technology
Micron Technology, Inc. is an industry leader in innovative memory and storage solutions that are transforming the way the world uses information to enrich everyone’s lives. With a consistent focus on customers, technology leadership, and manufacturing and operational excellence, Micron offers a rich portfolio of high-performance DRAM, NAND, and NOR memory and storage products. Every day, the innovations our people create power the data economy, enable advances in artificial intelligence (AI) and compute-intensive applications, and unlock opportunities from the data center to the intelligent edge to the client and mobile user experience. For more information about Micron Technology, Inc. (Nasdaq: MU), please visit micron.com.
© 2026 Micron Technology, Inc. All rights reserved. Information, products and specifications are subject to change without notice. Micron, the Micron logo, and all other Micron trademarks are property of Micron Technology, Inc. All other trademarks are the property of their respective owners.
Micron Product and Technical Communications Contacts:
Mengxi and Liu Guixuan
+1 (408) 444-2276
[email protected]
Micron Investor Information Contact:
Sathya Kumar
+1 (408) 450-6199
[email protected]
___________________________
1
https://epoch.ai/data-insights/context-windows
2
Trendforce 2026 data.
3
Based on Micron’s internal simulation predictions utilizing HBM4 288GB on a system with 1.5x more GPU FLOPs compared to current GPU systems with HBM3E 288GB.
4
One third of the power consumption is calculated based on the wattage of power used by one 128GB, 128-bit bus width SOCAMM2 module compared to two 64GB, 64-bit bus width DDR5 RDIMMs. The 1/3 footprint calculation compares SOCAMM2 area (14x90mm) to standard server RDIMMs.
5
Performance benefits are calculated comparing 9,200 MT/s and 6,400 MT/s products.
6
Rack space reduction is calculated as 720 drives per 36U x 245.76TB SSD for a total capacity of 176.9PB per rack. This compares to 720 drives per 36U x 44TB HDD for a total capacity of 31.7PB (theoretical maximum) per rack. The difference is that HDDs require 5.6 times more rack space.
7
The Micron 6600 ION 245TB SSD operates at 30W peak power, and each 44TB HDD operates at 10W peak power. Power information for 44TB HDD is not available. Comparison is based on peak power for 36TB/32TB HDD. Source: exos-ds2046.1-2512 en_us.pdf
8
Comparison of component pin speed specification levels between GDDR7 and GDDR6.
9
Improve your inference workloads with higher bandwidth in GDDR7.
10
The LLM used in testing was Llama 2 with 13 billion parameters and 10.4 GB file size.
11
Performance and power efficiency statements are based on a comparison of PCIe Gen4 Micron 3500 SSD and Micron 4600 SSD.
