Samsung Electronics Co. has reportedly reached a 60% yield rate in producing its next-generation mobile application processor (AP), the Exynos 2600, using its 2nm GAA process technology. The company aims to price the chip US$20 to US$30 lower than Qualcomm's Snapdragon equivalent.
AI has pushed the memory industry into a growth cycle. Etron Technology President Elvis Deng stated that memory supply is now in short supply across the board. DDR4, LPDDR4, and DDR3 shortages will be difficult to ease in the short term, and are expected to persist into the second half of 2026—2027.
Samsung's P5 plant at the second campus in Pyeongtaek, South Korea, will begin construction, with operations expected to start in 2028. Due to the weak memory market, P5 was previously put on hold. With the memory supercycle emerging, analysts state that the start of P5 construction means subsequent investment plans have also begun to take shape.
Samsung Electronics has been in active discussions with Samsung Electro-Mechanics (Semco), Japan's Ibiden, and Shinko Electric Industries regarding collaboration on semiconductor glass substrates. However, reports indicate that they have not yet advanced to formal sample evaluation because of Samsung's assessment of the technology and market potential of glass substrates.
When Dr. Wei-Jen Lo, the former senior vice president for corporate strategy development at TSMC, retired in July at the age of 75, many in the chip industry assumed his career had drawn to a close. Lo, an architect of several major process breakthroughs at TSMC and a key figure in the company's rise to technological dominance, was widely expected to honor strict noncompete commitments after stepping down.
Samsung Electronics has released its regulatory filing for the third quarter of 2025, revealing that Google—whose parent company is Alphabet—has once again reappeared on Samsung's list of top five customers. The company's report also formally incorporates the $16.5 billion chipmaking contract that Samsung was reported to have secured with Tesla in July.
Arm and Nvidia announced new steps to expand their collaboration on AI data center infrastructure, highlighting the growing demand for energy-efficient computing as AI workloads scale. Companies across the sector are prioritizing "intelligence per watt," as power constraints increasingly limit data center capacity and performance.
Dynamic Holding, a leading producer of automotive printed circuit boards (PCBs), is gearing up for a surge in artificial intelligence (AI) computing demand from application-specific integrated circuit (ASIC) customers anticipated in the second half of 2026. The company projects that AI-related revenues will exceed 20% of total sales for the first time, with contributions from both GPU and ASIC sectors balancing equally.
Orient Semiconductor Electronics (OSE) said the global memory industry has shifted decisively into a "seller's market," prompting the company to pass higher costs to customers and accelerate capacity expansion.
Microsoft CEO Satya Nadella recently stated in an interview that Microsoft can access OpenAI's model research outcomes. System-level innovations developed by OpenAI can also be fully leveraged by Microsoft, which can implement these designs first for OpenAI and then extend or adapt them to its own purposes. Going forward, Microsoft will be able to share OpenAI's proprietary chip-related technologies and use them as a foundation to adjust or expand its own application-specific integrated circuit (ASIC) chip development.
Taiwan's Academia Sinica has unveiled its quantum chip fabrication platform (QC-Fab) and quantum computing test platform (QC-Test), marking a major step in the nation's push into quantum technology. QC-Fab began trial operations in September 2025, while QC-Test is expected to open for trials in 2026. A 10-qubit quantum computer chip under in-house development is slated for deployment in the first half of 2026.
More coverage
