CONNECT WITH US
Dec 1
Google’s TPU victory widens China’s path to a post-Nvidia future
Google's success training the Gemini frontier model entirely on in-house TPUs has boosted confidence among China's tech giants. It strengthens the view that reducing Nvidia dependence is both viable and commercially sound, accelerating investment in domestic tensor-processor designs and other non-GPU accelerators.
Nvidia's US$2 billion investment in Synopsys marks a pivotal attempt to extend CUDA and GPU-accelerated computing from AI training into the core of industrial and semiconductor engineering, tightening its grip on the entire AI server ecosystem, as ASIC servers, particularly TPU-based, are challenging Nvidia.
Artificial intelligence (AI) is fueling a DRAM supercycle, prompting South Korean securities firms to continuously raise their forecasts for Samsung Electronics' operating profit in 2026. Some projections now approach KRW90-100 trillion (approx. US$61.3-68.1 billion).

Google CEO Sundar Pichai has dismissed concerns that the company's rapid expansion of its proprietary artificial intelligence infrastructure threatens Nvidia's market dominance. He argued that global demand for AI compute is growing fast enough for multiple chipmakers to prosper.

Although Taiwan possesses core upstream technologies in the robotics supply chain, downstream applications are only just getting started. To fill the gap, companies have formed the Robotics Innovation Alliance (RIA), emphasizing that future efforts will begin from actual industry needs to connect the supply and demand sides and push applications toward successful commercialization.
US AI company PaleBlueDot AI is reportedly seeking a US$300 million loan to help Chinese social media platform RedNote (Xiaohongshu) use Nvidia's high-end AI chips through a data center located in Tokyo.
Taiwan is experiencing the highest weekly average of cyberattacks in the Asia-Pacific region, distinguished by a notably large share of attacks targeting the "impact" phase, according to Fortinet's latest threat intelligence data. Derek Manky, Fortinet's global threat intelligence vice president, shared insights with DIGITIMES, revealing Taiwan's cybersecurity landscape deviates significantly from global patterns observed in the first three quarters of 2025.
A decade-old cooling idea returns as AI servers overheat
Dec 1
Rising power consumption in AI accelerators is pushing chipmakers to seek new thermal solutions. Jentech Precision general manager Lin Chin-lung said the company developed its Microchannel Lid a decade ago, yet the technology drew little interest until chipmakers began revisiting it over the past three years, with momentum accelerating in the last two. Customers are now focused on its two core benefits: substantially stronger heat dissipation and thinner system design, both critical for next-generation AI servers.
The rapid rise of artificial intelligence (AI) applications has significantly boosted demand for specialized glass-fiber fabrics in recent years, creating a growing supply shortage in the market. To address this trend,Formosa Plastics subsidiary Nan Ya Plastics and Japan's leading manufacturer Nitto Boseki (Nittobo) have announced a strategic partnership focused on specialized glass-fiber fabrics.
Shinkong Synthetic Fibers Corporation (SSFC) told investors on November 28 that it is navigating a challenging operating environment marked by raw material price volatility, Chinese product dumping, and shifting US tariff policies. The company said it is countering these pressures through diversified product development, rising contributions from its optoelectronics segment, and accelerating growth at subsidiary Shinsol Advanced Chemicals.
Optoelectronic semiconductor integration solution provider Brightek recently officially commissioned its Jiangsu factory, built with an investment of nearly CNY200 million (approx. US$28.3 million). Brightek stated that the new facility will fully support emerging applications such as international automotive manufacturers, smart mobility, intelligent sensing, and AI robotics, laying a critical foundation for growth over the next decade.
Silicon carbide (SiC) demand has been weighed down by a sluggish automotive market, but ACME recently observed signs of market reversal and expects the auto sector to gradually recover from late 2025 into 2026. The company is refocusing its R&D on SiC for AI data centers and inductive materials, while expanding production of semi-insulating grade SiC powder to meet surging defense industry needs.
AsiaAPE
biz
TSMC sues ex-VP who jumped to Intel
Will Taiwan, South Korea, Japan join semiconductor alliance with US?