Semiconductor News 20250924

Main News

1. NVIDIA and OpenAI form strategic alliance to build AI data center worth up to 15 trillion yen

NVIDIA and OpenAI have formed a strategic alliance and announced plans to invest up to 15 trillion yen ($100 billion) to build a massive AI data center. The plan includes integration of NVIDIA’s Vera Rubin AI platform, and NVIDIA has promised that despite the massive deal, GPU supply will not be affected and that all customers will be given top priority. This is one of the largest investments ever made in AI computing infrastructure and demonstrates NVIDIA’s commitment to reinvesting in its leading position in AI.

2. TSMC Increases N3P Prices by 20% Ahead of 2nm Era to Expand Foundry Market Dominance

TSMC is reportedly raising the price of its N3P process by 20% before the next generation 2nm era arrives, putting pressure on customers such as Qualcomm and MediaTek TSMC is further expanding its dominance in the foundry market, while Samsung is losing ground Samsung is losing ground. The CEO of Yole Group also stated that TSMC is leading the AI era in a major supply chain transformation, and that 15 companies, including Intel, AMD, and MediaTek, are among the first customers to adopt the 2nm process.

3.Samsung struggles with HBM4 yield; SK Hynix leads memory race

Samsung is reportedly struggling with HBM4 yield (yield), while SK Hynix is leading the high-bandwith memory (HBM) competition. Currently, Samsung is rushing to supply HBM3E to Nvidia, and the results of the Wafer Acceptance Test (WAT) are critically important for the supply to be realized. In contrast, SK Hynix is launching its M15 pilot line to produce 1 billion DRAMs to meet the surge in HBM demand from major tech companies.

4. NVIDIA and Intel Strengthen Strategic Alliance, Jointly Develop AI Superchip

NVIDIA is reportedly investing $5 billion in Intel to strengthen its alliance with Intel and jointly develop AI superchips; Intel joining Nvidia’s NVLink camp is described as having a significant impact on AI chip design, and the partnership is expected to boost long-term demand for ABF substrates. It is also suggested that this strong alliance will be a factor in further widening the technology gap between the U.S. and China.

5. Tesla Hires Semiconductor Talent for 2nm Samsung Collaboration and AI6 Chip Enhancement

Tesla is reportedly actively hiring experts in the field of semiconductor design in order to strengthen its in-house developed AI6 chip and to promote collaboration with Samsung in the 2nm process. This indicates that Tesla is accelerating its in-house manufacturing strategy by securing cutting-edge semiconductor technology in the areas of automated driving and AI computing.

6. MediaTek Considers U.S. Production at TSMC Arizona to Avoid Tariffs and Secure Customers

MediaTek is reportedly considering U.S. production at TSMC’s Arizona plant to secure customers while avoiding future tariff impacts. In addition, MediaTek’s new flagship chip Dimensity 9500, a significant upgrade to the ARM Mali-G1 Ultra, offers more FP32 computing power than the Xbox Series S and is targeted at the Apple A19 Pro to expand market share.

7. iPhone 17 Demand Surges; Apple Sees Opportunity to Capture Over 300 Million Users

Demand for the iPhone 17 is surging in the marketplace, and T-Mobile has achieved its largest iPhone weekend ever with the iPhone 17. for Apple, this represents a significant This is a huge business opportunity for Apple. Meanwhile, some iPhone 17 and 17 Pro users have reported Wi-Fi issues that appear to be caused by the iOS 26 update.

8. SK Hynix starts operation of M15 pilot line to produce 1 billion DRAMs to meet HBM demand

SK Hynix has reportedly launched its M15 pilot line for 1 billion DRAM production to meet the surge in demand for High Bandwidth Memory (HBM) from major tech companies. The company is following in the footsteps of Micron and Samsung in negotiating price adjustments with customers, a move that will increase its supply capacity and price dominance in the AI-era memory market.

Intel changes iGPU driver support structure for 11th-14th generation Core processors

Intel has announced that it has separated the built-in graphics (iGPU) drivers for its 11th through 14th generation Core processors into a separate lineage. This confirms that gaming support for these processor graphics will be discontinued on the first day of launch (Day 0). This indicates that Intel is focusing resources on new GPU products (such as Arc) and has revised its support strategy for traditional integrated graphics.

10. advanced packaging and photoelectric integration will be the focus of technology in the AI era

With the growing demands of AI and HPC (High Performance Computing), advanced packaging technologies are becoming increasingly important, and 12-inch SiC substrates are poised for a two-step introduction to CoWoS packaging. Advanced packaging was also shown to be a major focus at the SEMI event, as optoelectronic fusion startups invested by NVIDIA and TSMC are developing the technology with the goal of starting volume production in 2027. This move to integrate optical technology into chips aims to dramatically improve data transfer efficiency.

News List