Global semi news — Korea, China, Taiwan, the US, and Japan. Government policy, export controls, capex moves, supply-chain shifts, and macro events. AI-classified and tagged with affected tickers. All headlines link back to the originating publisher.
Original: AI 랠리에 균열… 미국 반도체·빅테크 이틀째 급락 지속 - 뉴스투데이
US semiconductor and big tech stocks extended sharp losses for a second consecutive session, signaling cracks in the AI-led rally. The pullback raises near-term concerns for Asian AI supply chain names including HBM and foundry leaders tied to NVIDIA and hyperscaler demand.
Why it matters: Sector-wide AI sentiment pullback rather than a specific policy or company event, but directly drags AI-exposed semi names globally.
Original: 오라클 — 증권 발행(424B5)
Securities offering filed 2026-06-23. See EDGAR for prospectus.
Why it matters: SEC 424B5 filing
Original: GPU 막히자 CPU로 우회…中, 슈퍼컴 1위로 美 제쳤다
China reportedly displaced the US at the top of the global supercomputing rankings using a CPU-centric architecture, circumventing US GPU export restrictions. The development underscores how export controls are accelerating China's domestic chip self-sufficiency in high-performance computing, with potential long-term implications for US semiconductor demand from Chinese HPC buyers.
Why it matters: Geopolitical export-control story with sector-wide implications for US HPC/GPU vendors but no direct earnings or product event for tracked names.
Open source articleOriginal: ＡＩ・半導体など成長分野の予算に「特別枠」、高市首相きょう表明…複数年度で計画・概算要求の上限設けず - 読売新聞
Japan's PM Takaichi will today announce a 'special frame' in the FY budget for growth sectors including AI and semiconductors, planned across multiple years with no cap on budget requests. The move signals sustained fiscal support for domestic chip ecosystem players (Rapidus, equipment/materials suppliers) and could expand subsidy access for foundry and advanced-node investments in Japan.
Why it matters: Direct Japanese policy signal for sustained semi subsidies benefits domestic equipment/materials makers and Rapidus ecosystem, but lacks specific allocation numbers and near-term earnings impact.
Original: TSMC cuts 28nm wafer starts by 25%, opening the door for UMC to grab market share - Crypto Briefing
TSMC is reportedly reducing 28nm wafer starts by 25%, signaling a strategic pullback from mature node capacity as it prioritizes leading-edge AI chip production. The move creates an opening for UMC and other mature-node foundries to capture displaced demand, with implications for 28nm pricing and utilization across the foundry sector.
Why it matters: TSMC mature-node capacity reallocation is a meaningful foundry sector signal affecting UMC and mature-node pricing, but not a near-term earnings or policy event.
Open source articleOriginal: 华为2026下半年新机时间表曝光!7款机逐月发布，一文看懂怎么选_旗舰_系列_Mate - 新浪网
Chinese media details Huawei's H2 2026 phone release schedule with 7 models launching monthly, including new Mate flagships powered by domestic Kirin chips. The cadence signals sustained Huawei smartphone momentum in China, reinforcing the domestic substitution narrative against foreign chipsets and pressuring Qualcomm/MediaTek share in the Chinese premium segment.
Why it matters: Huawei's sustained smartphone cadence with Kirin chips reinforces SMIC demand and pressures Qualcomm's China smartphone exposure, but the article is a release-schedule leak rather than a new tech or capacity disclosure.
Open source articleOriginal: 슈퍼마이크로, NVIDIA Vera Rubin NVL4 기반 엔드투엔드 DCBBS 청사진 공개
Supermicro launched a Data Center Building Block Solutions (DCBBS) roadmap centered on NVIDIA's Vera Rubin NVL4 platform, leveraging native FP64 performance to converge HPC and AI infrastructure workloads. The integrated rack-scale offering positions Supermicro to capture demand from hyperscalers and sovereign AI buildouts seeking unified HPC/AI deployments. NVIDIA's Vera Rubin family is the successor generation after Blackwell Ultra, signaling accelerated platform cadence.
Why it matters: NVIDIA next-gen platform productization signals accelerating AI infra cycle and CoWoS/HBM demand, though Supermicro itself is outside the tracked universe.
Open source articleOriginal: Auto Alliance’s John Bozzella on China and USMCA; DRAM chip crunch hits automakers - Automotive News
Auto Alliance CEO John Bozzella discusses China trade and USMCA as a DRAM shortage squeezes automakers, signaling tightening memory supply spilling beyond servers/AI into auto MCUs and modules. The crunch reinforces the ongoing DRAM upcycle narrative and pricing power for Korean memory suppliers.
Why it matters: Sector-wide DRAM tightness signal extending into automotive demand, supportive for Korean memory names but no company-specific event.
Open source articleOriginal: 中信建投：半导体设备全球景气周期持续确认 看好半导体产业趋势 - 新浪财经
China's CITIC Securities reiterates a bullish call on the semiconductor equipment cycle, citing sustained global capex momentum and structural tailwinds from domestic substitution in China. The note frames the upcycle as confirmation that mainland fabs (SMIC, CXMT, Huawei-linked) will continue driving WFE demand, with read-through to global equipment leaders AMAT, LRCX, KLAC and Asian peers, though Chinese sell-side typically positions this as accelerating self-sufficiency rather than benefiting US tools long-term.
Why it matters: Sell-side sector note on global WFE cycle with China domestic substitution angle — affects tracked equipment names but no new data point or policy catalyst.
Original: Micron Technology advances New York fab plan, boosting US AI memory capacity - Crypto Briefing
Micron is moving forward with its Clay, NY megafab project, reinforcing US-based DRAM/HBM supply for AI workloads. The advance signals continued execution on CHIPS Act-backed domestic memory capacity, with implications for HBM competitive dynamics versus Samsung and SK Hynix.
Why it matters: Fab buildout progress for a major HBM/DRAM supplier is a sector-relevant capacity signal but not a near-term earnings or policy event.
Cheonbo
278280
₩34,750
-7.33%