OpenAI and Oracle have announced plans to construct a 1GW data center campus in Saline Township, Michigan, with construction scheduled to begin in early 2026. The site, part of the Stargate Project, is expected to create over 2,500 construction jobs.

Amid the intensifying global artificial intelligence (AI) race, LG AI Research has expressed confidence that the risk of an AI bubble remains low, outlining plans to further transform corporate operations and R&D through AI agents. At the KES 2025, the company showcased its large language model "EXAONE," designed to enhance expert capabilities and enable general employees to perform at specialist levels. The model is being expanded across various industries through a growing network of partnerships.

As world leaders gather for the APEC Summit in Gyeongju amid tense geopolitics, South Korea is quietly taking on a new strategic role in the global AI race. Nvidia CEO Jensen Huang's landmark decision to allocate 260,000 of its latest Blackwell-architecture GPUs to Korea's top companies, including Samsung, SK, Hyundai Motor, and Naver Cloud, marks one of the most concentrated AI infrastructure deployments ever made in a single country.

The rapid expansion of the semiconductor industry and artificial intelligence (AI) supply chains is driving a sharp increase in global electricity demand, raising new challenges for power system resilience. Industry experts warn that as AI data centers scale to gigawatt (GW) levels, existing energy infrastructure may struggle to keep pace.

A team of researchers from National Taiwan University (NTU) has developed a technology that can precisely cut off generative AI models' associations with high-risk concepts. This innovation aims to make generative AI safer by preventing the generation of violent, deepfake, or unauthorized imitation content.

Ever since the launch of ChatGPT in 2022, market watchers have had high hopes for AI to transform industries by boosting profitability and operational efficiency. Terence Liao, general manager of Dell Taiwan, has identified 2025 as a watershed year for the widespread adoption of AI, pointing out that corporations can now accelerate as if on a flywheel by leveraging four main capabilities of AI.

The global AI race enters a new phase in 2025, shifting focus from advanced cloud algorithms and high-performance computing chips to the coldest yet most critical physical constraint: electricity. Analysis firm Goldman Sachs highlights that power provision is fast becoming the "physical limit" for AI development.

Rare earth elements (REEs), comprising 17 chemical elements including the 15 lanthanides from lanthanum (La, atomic number 57) to lutetium (Lu, 71), plus scandium and yttrium, are essential materials with unique optical, electrical, magnetic, and catalytic properties. Often called "industrial vitamins" or "industrial gold," REEs are indispensable in many high-tech industries.

At a seminar held on April 28, 2025, by Taiwan's Industrial Technology Research Institute (ITRI), experts highlighted the stark contrast in quantum talent development and industry application between Taiwan and the Netherlands. Jeff Lin, general director of ITRI's Industry, Science and Technology International Strategy Center, noted that although the Netherlands has a population of 18 million and produces about 200 quantum master's and PhD graduates annually, Taiwan—with 23 million people—only trains around 50 quantum specialists each year. Moreover, Taiwanese quantum talents often find it difficult to secure relevant jobs.

Samsung Electro-Mechanics has reported robust financial results for the third quarter of 2025 and projects sustained strong demand for multilayer ceramic capacitors (MLCCs) in 2026, driven by AI servers and automotive electronics. The company aims to achieve double-digit growth next year.