SK hynix 256 GB DDR5 module slashes Intel Xeon power consumption by 18%, saving data centers millions

2 Sources

Share

SK hynix announced its 256 GB DDR5 RDIMM based on 32Gb DRAM has passed Intel's certification for Xeon 6 platform, marking an industry first. The new server memory modules deliver 18% lower power consumption and 16% better inference performance, potentially saving hyperscale data centers millions of dollars through reduced energy costs.

SK hynix achieves industry-first Intel certification for 256 GB DDR5 memory

SK hynix announced that its 256 GB DDR5 RDIMM based on 32Gb DRAM has become the first memory module at this capacity to pass the Intel Data Center Certified process for the Intel Xeon 6 platform

1

. Testing and validation were completed at Intel's Advanced Data Center Development Laboratory in the United States, marking a significant milestone for server memory modules designed to handle demanding AI workloads

2

. This certification positions SK hynix to capture a larger share of the rapidly expanding server DDR5 DRAM market as data center operators seek high-capacity memory solutions that balance performance with energy efficiency.

Source: Tom's Hardware

Source: Tom's Hardware

Dramatic power efficiency gains through advanced 1b process technology

The new 256 GB DDR5 RDIMM achieves up to 18% lower power consumption compared to previous 256 GB products, thanks to SK hynix's 1b process technology

1

. A single 32Gb DRAM chip manufactured on the 5th Generation 10nm-class DRAM process consumes significantly less power than two 16Gb memory ICs made on the company's 1a production node. This translates to a 32W-per-socket reduction in a fully equipped 12-channel Intel Xeon memory subsystem, which typically draws between 180W and 300W depending on workloads

1

. For hyperscale data center operators running tens of thousands of machines, this power efficiency improvement represents substantial energy savings that could amount to millions of dollars annually.

Enhanced inference performance addresses growing AI server demands

Beyond power efficiency, servers equipped with the new 32Gb-based 256GB RDIMM deliver 16% better inference performance compared to 128GB modules using 32Gb DRAM

2

. This performance boost matters significantly as AI workloads continue to expand across cloud deployments. While AI servers consume premium high-bandwidth memory like HBM used on accelerators such as Nvidia B300, they also rely heavily on commodity server DDR5 SDRAM connected to x86 processors

1

. The combination of higher capacity, improved power efficiency, and stronger inference performance positions these modules as essential components for next-generation AI infrastructure.

Strategic implications for data center operators and memory market

Sangkwon Lee, head of DRAM Product Planning & Enablement at SK hynix, stated: "We are now able to respond more swiftly to customer needs, solidifying our leadership in the server DDR5 DRAM market. As a full-stack AI memory creator, we will actively address the growing demand for high-performance, low-power, and high-capacity memory solutions to further enhance customer satisfaction"

1

. Dimitrios Ziakas, vice president of Platform Architecture at Intel, added that the collaboration "has resulted in excellent results and contributed to the advancement of memory technology"

2

. The validation strengthens SK hynix's position to expand cooperation with major global data center operators as they upgrade infrastructure to support increasingly memory-intensive AI applications. Data center operators should monitor how these energy savings scale across their deployments, particularly as memory power consumption rivals CPU power draw in modern server configurations.

Today's Top Stories

TheOutpost.ai

Your Daily Dose of Curated AI News

Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Spend less time searching for the latest in AI and get straight to action.

© 2025 Triveous Technologies Private Limited
Instagram logo
LinkedIn logo