Artificial intelligence (AI) infrastructure is evolving faster than traditional server hardware cycles can accommodate. As AI workloads expand in scale and data intensity, memory architecture has become a key constraint in next-generation data center systems. In response, a new memory standard known as SOCAMM2 is gaining attention across the AI server ecosystem.

Lens Technology said SSDs assembled for enterprise NVMe storage supplier DERA have entered mass shipment at its facility in the Xiangtan Economic and Technological Development Zone, marking the company's expansion into the high-end data center storage supply chain.

Samsung Electronics is accelerating the expansion of its Pyeongtaek semiconductor campus in South Korea, aiming to establish the site as a core production base for next-generation AI memory. The company's fifth fabrication plant, P5, is scheduled to begin operations in 2028 and will support the long-term supply of high-bandwidth memory (HBM) and advanced DRAM used in AI servers and data-center infrastructure, according to Yonhap and Munhwa.

The US is moving to bar federal agencies from buying certain semiconductors tied to major China-based chipmakers, widening procurement restrictions even as memory shortages and rising prices strain electronics supply chains.

With the semiconductor industry entering a supercycle and reporting record earnings, Samsung Electronics and SK Hynix are preparing to launch large-scale semiconductor hiring.

Memory price increases are beginning to affect the server market. Wiwynn said that in 2025, server customers and memory suppliers were protected by long-term pricing agreements, but price growth accelerated sharply after the 2026 Lunar New Year. Customer demand remains unchanged. Higher memory costs will pressure ODM gross margins, while final profit and earnings per share (EPS) are expected to remain stable.