July 23, 2025

Information collected & summarized by:

Intel Earnings Spotlight Contract Chipmaking Shift

Intel is poised to report its sixth straight quarterly loss and a fifth consecutive revenue decline on Thursday, thrusting new CEO Lip Bu-Tan’s contract manufacturing ambitions into the spotlight.

Investors will zero in on Intel’s strategy to pivot from the in-house 18A process to a next-gen 14A node aimed at winning external foundry customers, a move that could trigger multibillion-dollar writedowns. Analysts expect a net loss of about $1.25 billion on roughly $11.9 billion in sales, while the nascent foundry unit seeks to reach breakeven by 2027.

Key points:

Intel sold 51% of its Altera business for $4.46 billion in April.

Foundry sales are forecast at $4.49 billion in Q2, mostly internal volumes.

PC unit revenue may dip 2%, but data-center sales could rise ~20%.

Taiwan Unveils $510 Billion AI Infrastructure Plan to Bolster Chip Industry

Taiwan’s government has launched its Ten Major AI Infrastructure Projects, pledging over NT$15 trillion (about $510 billion) by 2040 to cement its role as an AI and semiconductor powerhouse. The plan zeroes in on silicon photonics, quantum technology and AI robotics, and sets up the AI Robotics Industry Grand Alliance with Foxconn’s backing.

To accelerate innovation, Taipei will channel more than NT$100 billion ($3.1 billion) into AI-focused venture capital, target 500,000 new jobs and establish three world-class research labs—all while aiming for balanced regional development and supply-chain resilience.

Trending Newsletters in Nail

Capitol Brief

368 new followers today

Global Breaking News Digest

306 new followers today

California State News

312 new followers today

View More Topics

CXMT Pushes DDR5 Production to Late 2025

China’s state‐backed DRAM maker ChangXin Memory Technologies (CXMT) has pushed its DDR5 chip rollout to late 2025, citing ongoing yield and reliability challenges despite recent quality gains. Early samples suffered stability issues at both high (60 °C) and sub‐zero temperatures, forcing a costly photomask redesign and postponing initial mass‐production plans slated for mid‐2025.

Although CXMT says module performance now matches Taiwan’s Nanya, its yields remain just above 50%, well below commodity DRAM standards. With U.S. export curbs limiting maintenance of key fab tools in China, the company’s aggressive capacity expansion—from 170,000 to around 280,000 300 mm wafers per month—may also face headwinds.

Only Public Hearing Set for Micron’s Clay Fab Impact Report

Micron Technology’s long-awaited draft environmental impact statement (EIS) for its proposed Clay, New York, chipmaking campus finally faces public scrutiny on Thursday, when a day-long hearing opens at Liverpool High School. This single session, running from 10 a.m. to 9 p.m. with two breaks, will be the only opportunity for residents to voice support or concerns in person on the 20,000-page report.

Registration begins at 9:30 a.m., and speakers are expected to observe the customary 2–3 minute limit. The Onondaga County Industrial Development Agency and U.S. Department of Commerce will listen but not respond, then incorporate all substantive remarks into a final EIS due in October. Written comments remain open through August 11.

Reconfigurable Memristor System Enables In-Memory Data Sorting

Peking University researchers have unveiled a reconfigurable memristor-based system that performs fast, energy-efficient in-memory data sorting—a task long considered a stumbling block for compute-in-memory (CIM) architectures. Led by Professor Yuchao Yang and detailed in Nature Electronics (DOI: 10.1038/s41928-025-01405-2), the new design integrates one-transistor-one-resistor (1T1R) memristor arrays with three reconfigurable peripheral modules.

Key innovations include support for multiple data types (unsigned, two’s-complement, floating-point) and three parallelism enhancement strategies:

Multi-bank: Distributes data across banks for level parallelism

Bit-slice: Processes bitwise segments in parallel

Multi-level: Exploits in-device conductance levels

Initial tests demonstrate significant energy savings versus prior approaches, paving the way for broader CIM adoption in AI training, robotics, and smart systems.

Newsletter you might like

Electric Drive

Stay ahead of the curve with the latest EV technology, market trends, and sustainable driving news.

LLM Digest

Stay ahead of the curve with the latest news about Large Language Models

Electric Drive

Stay ahead of the curve with the latest EV technology, market trends, and sustainable driving news.

LLM Digest

Stay ahead of the curve with the latest news about Large Language Models

Explore More

Unsubscribe

Terms & Conditions

Taiwan Launches $510 B AI Infrastructure Plan to Boost Semiconductor Leadership