惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

博客园 - Franky
D
Docker
Jina AI
Jina AI
The GitHub Blog
The GitHub Blog
博客园 - 聂微东
B
Blog RSS Feed
大猫的无限游戏
大猫的无限游戏
M
MIT News - Artificial intelligence
Vercel News
Vercel News
Microsoft Security Blog
Microsoft Security Blog
博客园 - 叶小钗
爱范儿
爱范儿
D
DataBreaches.Net
Hugging Face - Blog
Hugging Face - Blog
IT之家
IT之家
Recent Announcements
Recent Announcements
U
Unit 42
腾讯CDC
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
宝玉的分享
宝玉的分享
量子位
Stack Overflow Blog
Stack Overflow Blog
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
Microsoft Azure Blog
Microsoft Azure Blog

TechWire Asia

Microsoft to deploy AMD Helios AI systems on Azure Microsoft ERP Implementation Partners Roundup: Which Is Best For Your Business? Semiconductor talent retention now comes with a price tag, and it keeps climbing AI token usage is not productivity. Here's what to measure instead Oracle brings governed AI agents to Fusion Applications Nvidia expands Japan AI infrastructure and robotics push AI Appreciation Day 2026 puts trust and governance in focus NVIDIA pours its full stack into Japan. The flip side of its China lockout? Malaysia's digital regulations are becoming a real cost for its startups Malaysia's AI data center vision: How EdgeConneX is building for the future Southeast Asia tech funding doubled to $7.4 billion. One company took most of it SK Hynix's Nasdaq listing raises $26.5 billion to fund Korea's AI memory expansion OpenAI launches GPT-5.6 for coding, cyber and science Meta rolls out Muse Image AI model for Instagram, WhatsApp, and advertisers Malaysia businesses face AI and password cybersecurity risks How AI workloads will test APAC mobile networks Enterprise AI costs don't have to spiral, argues ManageEngine Microsoft launches $2.5B Frontier Company for enterprise AI FIFA World Cup: How To Win Fans in APAC With Technology Kanga enters a new phase of global growth and launches Kanga Global Vertiv ramps up manufacturing in Johor's tightening data centre market U Mobile completes migration to own ULTRA5G network after DNB exit Anthropic Claude models launch in Microsoft Foundry on Azure Asia built the AI infrastructure boom. The BIS just flagged who's exposed if it stalls. Why Apple is lobbying Washington to buy China’s memory chips Nvidia-backed Firmus plans 170,000-GPU Batam AI data centre Taiwan robot makers march into humanoid systems IBM claims world’s first sub-1 nm chip technology using nanostack design Can Alibaba bridge Malaysia’s SME talent gap via agentic AI for business? Huawei’s new tech explains why mobile AI network tech is no longer optional
Memory shortage to persist through 2027 as supply lags de...
Muhammad Zul · 2026-04-20 · via TechWire Asia
  • Memory shortage to persist through 2027 as supply lags demand.
  • AI demand and limited capacity keep memory prices high and shipments weak.

Rising memory costs are expected to persist through at least 2027, as supply continues to lag demand, according to Nikkei Asia. Leading US and South Korean suppliers are increasing DRAM output at a pace that would meet only about 60% of demand by 2027.

The report estimates that memory production would need to expand by around 12% annually through 2027 to meet demand. Current growth stands at approximately 7.5%, while a gap of about 40% remains between supply and demand.

Supply constraints persist

Additional data from market research firms indicates that price increases have accelerated in recent quarters. TrendForce estimates that DRAM contract prices rose by approximately 90% to 95% quarter over quarter in early 2026, while NAND flash prices increased by 55% to 60% over the same period.

TrendForce said persistent demand from AI servers and data centres in the first quarter of 2026 further worsened the supply-demand imbalance, increasing suppliers’ pricing power.

AI demand and memory allocation

Nikkei Asia also reported that demand for memory has increased with AI-related workloads, which require higher volumes of DRAM. Samsung Electronics, SK Hynix, and Micron Technology together account for about 90% of the global DRAM market and are effectively the only companies able to produce high-bandwidth memory at scale.

TrendForce and IDC have also pointed to limited capacity expansion following the previous industry downturn as a factor constraining supply, reflecting the time required to bring new fabrication capacity online.

In recent years, those suppliers have prioritised high-bandwidth memory, Nikkei Asia reported. Production increases for general-purpose memory used in PCs and smartphones were delayed, contributing to shortages from late 2025.

Manufacturing capacity expansion is underway but remains limited in the near term. Samsung Electronics plans to bring a fourth fabrication plant at its Pyeongtaek campus online in 2026. However, full-scale mass production at the facility is not expected until 2027 or later.

The Pyeongtaek site is also designed to produce logic chips, which will constrain its ability to increase memory output. Samsung is also developing a fifth production line at the same location focused on AI-oriented memory, though this facility is not expected to begin operations until 2028 or later.

Samsung memory chief Kim Jae-june said in January that production scale remains constrained in the industry and that supply increases in 2026 and 2027 will be limited.

High-bandwidth memory, or HBM, is a type of high-performance DRAM used in AI semiconductors. SK Group Chairman Chey Tae-won said supply constraints in AI memory could continue into 2030 due to wafer shortages and limits on scaling production.

In the interim, SK Hynix began operating an HBM plant in Cheongju in February. According to Nikkei Asia, it is one of the few production increases among leading suppliers expected to add to supply in 2026.

SK Hynix is also accelerating construction of a plant in Yongin, with completion targeted for February 2027, three months earlier than originally planned. Micron, meanwhile, plans to begin HBM production in Idaho and Singapore in 2027, and aims to start mass production at a new Hiroshima facility in 2028.

Impact on devices and supply chains

Memory demand remains concentrated in a small number of product categories. About 80% to 90% of memory chips are used in PCs and servers, with the remainder going to automobiles, industrial equipment, and other products, Nikkei Asia reported.

The pressure is already showing up in end markets. IDC said worldsmartphone shipments are expected to fall 12.9% in 2026 to 1.12 billion units, while the PC market is projected to decline 11.3% in 2026.

Cost pressures are more visible in lower-priced devices. Memory accounts for about 20% of the manufacturing cost of low-priced smartphones and could reach nearly 40% by mid-2026, according to IDC figures cited by Nikkei Asia.

TrendForce reported that rising NAND flash prices, with DRAM increases, are contributing to higher overall component costs in PCs and other devices. Its March outlook said DRAM suppliers continue reallocating capacity toward server-related applications, while NAND capacity is increasingly being directed to enterprise SSDs.

Supply chain conditions remain a contributing factor to pricing pressure. Disruptions linked to tensions around the Strait of Hormuz have affected shipping routes and energy markets. Reuters reported that the strait typically carries about 20% of the world’s oil and liquefied natural gas flows.

Nikkei Asia also reported that turmoil in the Middle East is increasing electricity and materials costs, adding further uncertainty to the production outlook.

Counterpoint Research said supply and demand conditions in the memory market are not expected to stabilise until around 2028 based on current production and demand trends.

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events, click here for more information.

AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.