CONNECT WITH US
Jul 9, 12:20
IBM taps Samsung's 5nm process for Spyre AI chip in new Power11 servers
IBM has introduced its latest Power11 enterprise servers in Germany, featuring the Spyre accelerator chip designed for AI inference. The chip is manufactured using Samsung Electronics' 5nm process and is integrated into various server models aimed at multiple industries, including finance, healthcare, government, and manufacturing.
India-based EMS provider Kaynes Technology is expected to open a new manufacturing unit in Kerala. The company will receive land in Perumbavoor Industrial Park, developed by KINFRA, following approval from State Industries Minister P. Rajeeve after a site visit and discussions.
Sysgration posted NT$282 million (approx. US$9.7 million) in consolidated revenue for June 2025, marking year-over-year and month-over-month improvements of 12.83% and 5.65% respectively, as new products and clients begin contributing to revenue. For the second quarter of 2025, revenue grew by 9.50% year-over-year to NT$833 million (approx. US$28.6 million), while for the first half of the year, total revenue reached NT$1.706 billion (approx. US$58.6 million), increasing by 3.32% year-over-year even amid exchange rate fluctuations and other adverse external factors.
GSEO expects 2H25 orders to outpace 2H24
Jul 9, 12:36
Taiwanese lens maker Genius Electronic Optical (GSEO) announced its latest financial report on July 8 and expects revenue in the third quarter will grow compared with the second quarter, as it moves into the peak season in July.
Luxshare Precision, Lens Technology, and Goertek, China-based top-tier suppliers of Apple, are racing toward IPOs on the Hong Kong Stock Exchange (HKEX) as they seek capital to fuel transformation amid US-China trade frictions and supply chain realignments. Long defined by iPhone-related business, the trio is now banking on public listings to fund their next chapter in global manufacturing.
Taiwan-based Kinko Optical has strategically avoided low-price competition by expanding its mid- to high-end lens and next-generation product offerings, resulting in nearly 25% revenue growth in the first half of 2025.
US President Donald Trump notified 14 trading partners on July 7, 2025, of updated reciprocal tariff rates while postponing the original July 9 implementation date to August 1. The delay provides governments additional time to negotiate or adjust to US trade measures, with global shipping rates already showing sharp declines.
US President Donald Trump notified trade partners on July 7, 2025, Eastern Time, of new tariff rates set to take effect on August 1. He first informed Japan and South Korea that a 25% reciprocal tariff would be imposed on their goods, warning that rates could increase further if retaliatory measures are taken.
GlobalFoundries has announced an agreement to acquire MIPS, a supplier of AI and processor intellectual property (IP). The acquisition aims to broaden GlobalFoundries' customizable IP portfolio and enhance its process technologies with integrated IP and software capabilities.
With volume production of server chassis for Nvidia's GB200 turning stable, the operation of suppliers for related components such as cooling modules, sliding rails, and battery backup units (BBUs) has all turned positive, prompting them to hold a positive view for the second half of 2025. In addition, Nvidia's upcoming next-generation GB300 will help prolong the momentum during the period. Despite the requirement of a learning curve, industry visibility remains high, insulating its performance from market fluctuations.
Hans Dekkers, General Manager of IBM Asia Pacific, recently returned to Taiwan for his fifth visit in less than a year since taking office. He believes Taiwan and Japan will play important roles in the quantum field after reviewing Taiwan's notable growth in industry and academia surrounding quantum technology.

Groq, the US-based AI chipmaker, has launched its first European data center in Helsinki, accelerating its global expansion and tapping into the rising demand for sovereign AI infrastructure. Developed in partnership with Equinix, the facility will host Groq's proprietary Language Processing Units (LPUs), engineered for fast deployment and cost-efficient AI inference at scale.