Semiconductor News 20250827

Main News

1. NVIDIA Launches Jetson AGX Thor for Robotics and Physical AI, Accelerating Humanoid Robot Development NVIDIA has launched the Jetson AGX Thor, a Blackwell-powered embedded AI supercomputer to accelerate the adoption of robotics and “physical AI. NVIDIA has launched Jetson AGX Thor, a Blackwell-powered embedded AI supercomputer, to accelerate the spread of robotics and “physical AI. This is the first full-scale effort to give humanoid robots their own “brains” to make them smarter and more versatile, and several partners, including Infineon, are helping develop on this platform. nVidia aims to be “No. 1 Robot without Robots” and is driving the convergence of AI and robotics integration.
2. Huawei Launches AI SSD in Shanghai to Solve HBM Supply Bottleneck Huawei has announced its intention to enter the global AI memory competition by launching an AI SSD (Solid State Drive) in Shanghai to solve the supply bottleneck of HBM (high bandwidth memory). This is an important step in the development of AI. This is seen as an important step for Huawei to gain independence and competitiveness within the semiconductor ecosystem as memory becomes increasingly important in AI development. it aims to meet the demand for highly efficient data processing in AI computing.
3. Samsung and NVIDIA CEOs Hint at HBM4 Deal, Backed by $150 Billion in U.S. Support Samsung and NVIDIA CEOs met in Seoul to hint at a partnership on High Bandwidth Memory 4 (HBM4), a next-generation memory technology. This comes amid the South Korean government’s commitment to invest an additional $150 billion in the U.S. The meeting brought together industry leaders, dubbed the “Semiconductor Avengers,” to discuss future semiconductor strategy and cooperation. Cooperation between the two companies could have a significant impact on the high-performance memory market for AI.
4. China Report edly Imposes Full Export Halt on NVIDIA’s H20 Chips, Concerns AI Revenue The Chinese government has reportedly imposed a full sales halt on NVIDIA’s H20 AI chips. This action could cause serious uncertainty for NVIDIA’s AI-related revenues in China, highlighting the extent to which technology friction between the U.S. and China is impacting the cutting-edge semiconductor market. Geopolitical tensions in the global AI chip supply chain continue to mount.
5. AMD Unveils Details of Instinct MI350 GPU, a Strong Contender in the AI Hardware Market AMD has unveiled details of its cutting-edge AI hardware, the Instinct MI350 GPU. The GPU is a 3nm process 3D chiplet design based on the CDNA 4 architecture, with 185 billion transistors, 1400W thermal design power (TBP), and a massive 288GB of memory supporting over 4000B LLMs (large language models). This makes it a strong competitor to NVIDIA’s AI chips and is expected to increase competition in the AI chip market.
6. TSMC Report edly Eliminates Chinese Equipment from 2nm Fabs under U.S. Pressure TSMC, the world’s largest foundry, has reportedly eliminated Chinese manufacturing equipment from its 2nm process fabs under pressure from the United States. This indicates that the technology hegemony struggle between the U.S. and China is having a direct impact on advanced semiconductor manufacturing processes and their supply chains, and TSMC is likely responding to geopolitical imperatives and strengthening its supply chain risk management.
7. AMD and IBM Team Up to Develop “Quant um-Centric” Supercomputing AMD and IBM announced today that they are collaborating to develop “quantum-centric” supercomputing to usher in a new era of computation. This strategic alliance integrates quantum computing with traditional high-performance computing (HPC) in the pursuit of next-generation computational power. The goal is to create an advanced computing infrastructure for solving the complex scientific and industrial challenges of the future.
8. Rapidus Utilizes AI as an Essential Tool for 2nm Design and Announces Strategic Collaboration with Keysight Rapidus, a Japanese company responsible for next-generation semiconductor manufacturing, has introduced AI as an essential tool for 2nm process design and has proven its effectiveness. Rapidus also announced a strategic collaboration with Keysight, adopting Keysight’s analysis solution to improve the accuracy of its PDK (process design kit). This represents an important move by the Japanese semiconductor industry to leverage cutting-edge technology and AI to strengthen its global competitiveness.
9. Stanford Study Finds AI Dominating Entry-Level Programming and Customer Service Jobs A new Stanford University study finds that AI is rapidly dominating entry-level programming and customer service jobs. Over the past three years, job ads in these fields have declined by 13%, highlighting the potential impact of AI on the labor market and the increasing automation, especially in entry-level roles. This trend has important implications for future education and career path formation.
10. SK hynix Begins Mass Production of 321-Layer 2TB QLC NAND, Plans Product Launch in H1 2026 SK hynix has begun mass production of 321-layer 2TB QLC (Quad-Level Cell) NAND flash memory. This is a significant advance in memory technology, and products incorporating this technology are planned for market launch in the first half of 2026. This high-density, high-capacity NAND flash is expected to meet the demands of applications requiring massive data processing, such as data centers, enterprise storage, and AI workloads.

News List