用于人工智能的存储器推动着生产设备的投资。

Les mémoires pour l’IA propulsent les investissements en équipements de production

VIPress.net by Pascal Coutance 2026-07-08 16:06 Original
摘要
美国半导体协会Semi预测,受AI驱动的HBM(高带宽内存)和DDR5内存需求推动,全球用于存储器的300毫米晶圆制造设备支出将在2026年超过500亿美元。这项投资激增反映了AI应用对先进存储芯片的巨大拉动效应,将深刻影响半导体设备市场格局。

根据美国半导体行业协会SEMI的最新报告,受人工智能驱动的高带宽内存(HBM)与DDR5内存需求拉动,全球用于300毫米晶圆存储芯片生产的设备支出将在2026年突破500亿美元。这一投资浪潮凸显了AI应用对先进存储技术的强劲依赖,HBM因其高数据吞吐能力成为大规模AI训练和推理芯片的关键组件,而DDR5则在高性能计算和数据中心领域快速普及。SEMI的预测标志着存储器领域正经历结构性升级,设备制造商将迎来新一轮资本开支密集期,产能扩张重点集中在可满足严苛AI负载的尖端存储制程。

Summary
According to industry association SEMI, global spending on 300mm wafer fabrication equipment dedicated to memory chips will exceed $50 billion by 2026, driven by AI-fueled demand for high-bandwidth memory (HBM) and DDR5. This surge underscores the critical role of advanced memory technologies in enabling AI applications and is accelerating investments in semiconductor manufacturing equipment.

Global spending on 300mm wafer fab equipment dedicated to memory production will exceed $50 billion as early as 2026, propelled by soaring demand for HBM and DDR5 memories to power AI workloads, according to industry association Semi. The broader 300mm semiconductor equipment investment cycle is accelerating as chipmakers race to expand capacity for advanced memory technologies critical to generative AI, high-performance computing, and data centers.

High-bandwidth memory (HBM) stacks and next-generation DDR5 DRAM require increasingly complex manufacturing processes, driving up capital intensity per wafer. Semi’s forecast points to a sharp inflection: memory-related 300mm equipment expenditures, which had dipped during the last cyclical downturn, are set to nearly double from 2024 levels, surpassing the $50 billion mark within two years. This surge reflects concentrated investments by major memory manufacturers in new fabs and line conversions optimized for HBM3/E and high-capacity DDR5 modules.

The AI-driven memory boom is reshaping the equipment mix, with lithography, etching, and advanced packaging tools seeing disproportionate order growth. South Korea, the US, and China are expected to account for the bulk of the spending, underpinned by government incentives and strategic supply chain localization efforts. Semi’s data underscores how memory—long treated as a cyclical commodity—is now structurally tied to the secular AI infrastructure build-out, sustaining elevated capex through the decade’s end.

Résumé
Selon l’organisation américaine Semi, les dépenses mondiales en équipements de production de mémoires sur tranches de 300 mm dépasseront 50 milliards de dollars dès 2026, stimulées par la demande en mémoires HBM et DDR5 pour l’IA. Cette projection souligne l’impact massif de l’intelligence artificielle sur les investissements dans la fabrication de semi-conducteurs.

Les dépenses mondiales en équipements de production de semi-conducteurs en tranches de 300 mm dédiés aux mémoires dépasseront 50 milliards de dollars dès 2026, portées par la demande en mémoires HBM et DDR5 liée à l’IA, selon l’organisation américaine Semi. Les investissements mondiaux dans les équipements de production de semi-conducteurs en tranches de 300 mm […]

Cet article Les mémoires pour l’IA propulsent les investissements en équipements de production a été publié par VIPress.net.

AI Insight
Core Point

Global spending on 300mm wafer fabrication equipment dedicated to memory chips will exceed $50 billion by 2026, driven by AI-fueled demand for HBM and DDR5 memory, per industry body Semi.

Key Players
  • Semi — Semiconductor industry association providing market forecasts, based in USA.
Industry Impact
  • ICT: High — Record equipment investment signals a structural shift in memory manufacturing scale and supply chains.
  • Computing/AI: High — AI workloads are directly accelerating the need for high-bandwidth memory, reshaping capital allocation.
Tracking

[Strongly track] — The projected >$50B in memory fab equipment spending by 2026 marks a pivotal, AI-driven capital cycle that will redefine equipment vendor revenues and memory market concentration.

Related Companies
neutral
Categories
半导体
AI Processing
2026-07-08 17:17
deepseek / deepseek-v4-pro